METHOD FOR ESTIMATING THE ROTOR TORQUES OF AN AIRCRAFT CAPABLE OF HOVERING AND CONTROL UNIT FOR AN AIRCRAFT CAPABLE OF HOVERING Cross-Reference to Related Applications This Patent Application claims priority from European Patent Application No. 21217859.4 filed on December 27, 2021, the entire disclosure of which is incorporated herein by reference. Technical Field The present invention relates to a method for estimating the rotor torques of an aircraft capable of hovering. The present invention also relates to a control unit for an aircraft capable of hovering. Background Aircraft capable of hovering, such as convertiplanes or helicopters, are known comprising: - at least two rotors, which are operatively connected to each other and controllable independently of each other; - at least one engine, which is operatively connected to the rotors and is adapted to provide them with an engine torque; and - transmission units operatively connecting the engine to the rotors. In particular, the rotors are driven in rotation by respective rotor torques, which are in general different from the engine torque for several reasons. First, a portion of the engine torque is lost because of mechanical losses in the transmission units. In addition, a portion of the engine torque may be used to drive one or more accessories of the aircraft. The known aircraft further comprise physical sensors arranged at the two rotors and configured to directly measure the rotor torques. However, the rotor torques measured by such sensors (e.g., strain-based torque sensors) are not consistently reliable. In addition, the use of these physical sensors in aircraft results in increased installation and maintenance costs, as well as increased weight and complexity of the aircraft. This is especially relevant for the convertiplanes, the rotors of which are known to be tiltable relative to a reference system fixed with respect to stationary parts of the convertiplanes. Indeed, each time the rotors of the convertiplane are tilted, the physical sensors are tilted integrally with the rotors. This complicates the tilting movement of the rotors and the electrical connection of the sensors to the fixed parts of the convertiplane. Methods have therefore been developed to estimate the rotor torques acting on each of the two rotors without the physical sensors. In particular, the methods include the steps of: - subtracting from the engine torque the mechanical losses due to the transmission of the engine torque from the engine to the rotors; and - subtracting from the engine torque the mechanical energy supplied by the engine to any accessories of the aircraft connected thereto. The result of these subtractions is the total available engine torque. The rotor torque is then estimated by dividing the total available engine torque by the number of rotors of the aircraft. In other words, the torque acting on each rotor is estimated by apportioning the total available engine torque produced by the engine equally between the two rotors. Such an estimate can be considered sufficiently accurate under the assumption that the rotor torques acting on the two rotors are equal or substantially equal to each other. However, the rotor torques of two independently controllable rotors may actually be significantly different from each other. For example, this may occur during particular manoeuvres of the aircraft. As a result, the rotor torque calculated by the known estimation methods fails to give an adequate indication of the value of the torque acting on each of the two rotors, when the assumption that the rotor torques acting on the two rotors are equal or substantially equal to each other is not valid. US-A1-2017336809 discloses amethod for executing yaw control of an aircraft including two rotors. The method includes inducing helicopter yaw by creating a differential torque between the two rotors, wherein the creating of the differential torque comprises inducing a differential collective pitch to generate a differential thrust, and maintaining helicopter roll equilibrium during the inducing of the helicopter yaw by inducing a differential cyclic pitch to generate a differential lift offset. US-A1-2016122039 discloses a method for calculating torque through a rotor mast of a propulsion system of a tiltrotor aircraft including receiving the torque applied through a quill shaft of the rotorcraft. The quill shaft is located between a fixed gearbox and a spindle gearbox, and the spindle gearbox is rotatable about a conversion access. The torque through the rotor mast is determined by using the torque through the quill shaft and the efficiency loss value between the quill shaft and the rotor mast. In addition, a control method for turboshaft engine is disclosed in the paper “A novel control method for turboshaft engine with variable rotor speed based on the Ngdot estimator through LQG/LTR and rotor predicted torque feedforward” by WANG YONG et al. (ISSN: 1000-9361). Therefore, a need is felt within the sector to improve the estimation of the rotor torques of an aircraft comprising at least two independently controllable and operatively connected rotors, without using any physical sensor to directly measure the rotor torques. Summary The object of the present invention is to realise a method for estimating rotor torques, which allows the aforesaid need to be satisfied in a simple and efficient way. According to the invention, the aforesaid object is achieved by a method for estimating rotor torques according to claim 1. The present invention also relates to a control unit according to claim 8 and to an aircraft capable of hovering according to claim 15. Brief Description of the Drawings Three embodiments are described below for a better understanding of the present invention, provided by way of non-limiting example with reference to the accompanying drawings, wherein: - Figure 1 shows a top view of a first embodiment of an aircraft capable of hovering, in particular a convertiplane, comprising two rotors; - Figure 2 is a block diagram related to a method according to the present invention for estimating the torques of the aircraft of Figure 1; - Figure 3 is a detail of the block diagram of Figure 2; - Figure 4 is a schematic representation of some components of the aircraft of Figure 1, with parts removed for clarity; - Figure 5 shows a top view of a second embodiment of an aircraft capable of hovering comprising four rotors; - Figure 6 is a block diagram related to the method according to the present invention for estimating the torques of the aircraft of Figure 5; - Figure 7 is a schematic representation of some components of the aircraft of Figure 5, with parts removed for clarity; and - Figure 8 is a block diagram related to the method according to the present invention for estimating the torques of a third embodiment of an aircraft capable of hovering comprising any number of rotors. Description of Embodiments With reference to Figure 1, number 1 denotes an aircraft capable of hovering, in the shown case a convertiplane. It should be noted that the terms "front”, "longitudinal", "lateral", "above" and "below" and the like used in this description refer to a normal direction of advancement of convertiplane 1. The convertiplane 1 essentially comprises: - a fuselage 2 having an axis A of longitudinal extension; - a pair of half-wings 3 extending cantilevered from respective parts opposite one another of fuselage 2 and transversely to axis A; - a pair of nacelles 11, 12 housing at least partially respective engines 21, 22 and attached to respective half-wings 3; - a pair of rotors 31, 32 operatively connected with the respective engines 21, 22; and - a control unit 5. In detail, control unit 5 comprises in a known manner a computational unit, a memory, and one or more interface units for the electrical and operational connection to sensor means 40, 41, 42, 43 of aircraft 1, which will be described in detail in the following (Figure 4). Each rotor 31, 32 essentially comprises a hub 7 and a plurality of blades 8. Convertiplane 1 further comprises two transmission units 6, which are each operatively connected to a respective engine 21, 22 and the relative hub 7, to drive it in rotation. In particular, rotors 31, 32 are rotatable about respective axes B with respective angular speeds ω1, ω2. Preferably, angular speeds ω1, ω2 are constant over time. Furthermore, blades 8 are articulated on the respective hub 7 in such a way that respective angles of attack θ1 _COLL , θ2 _COLL are collectively adjustable relative to respective axes B. In the industry, angles of attack θ1 _COLL , θ2 _COLL are commonly known as "collective pitch angles" and will therefore be referred to that way in the following. Rotors 31, 32 are controllable independently of each other. In particular, the collective pitch angles θ1 _COLL and θ2 _COLL may be different from each other. More specifically, as the collective pitch angles θ1 _COLL , θ2 _COLL of one rotor 31, 32 increases, the thrust exerted by the rotor 31, 32 parallel to axis B and the drag torque acting on the rotor 31, 32 correspondingly increases. As a result, angular speeds ω1, ω2 remain substantially constant. Furthermore, rotors 31, 32 are identical to each other. In detail, rotors 31 and 32 have the same power required for the same input. This means, in further detail, that for a given collective input and a given boundary condition, the torque required to rotate rotors 31 and 32 is the same or substantially the same, for all inputs and operating conditions. Convertiplane 1 further comprises sensor means 40 configured to measure collective pitch angles θ1 _COLL , θ2 _COLL of respective rotors 31, 32. Sensor means 40 are configured to generate a signal associated with collective pitch angles θ1 _COLL , θ2 _COLL and are operatively connected to control unit 5 (Figure 4). Furthermore, rotors 31, 32 are tiltable with respect to respective axes C relative to half-wings 3. In particular, axes C are transverse to axis A and axes B (Figure 1). It is important to note that axes A, B and C are fixed with respect to convertiplane 1. Therefore, convertiplane 1 can be selectively arranged: - in a “helicopter” configuration (not shown), wherein axes B of rotors 31, 32 are orthogonal to axis A and parallel to axes C; and - in an "airplane" configuration (shown in Figure 1), wherein axes B of rotors 31, 32 are parallel to axis A and orthogonal to axes C. In detail, it is possible to define angles α1, α2 corresponding respectively to the orientation of rotational axes B with respect to axis A. Convertiplane 1 further comprises sensor means 41 configured to measure angles α1, α2 of respective rotational axes B of rotors 31, 32 with respect to axis A. In detail, sensor means 41 are configured to generate a signal associated with angles α1, α2 and are operatively connected to control unit 5 (Figure 4). In the embodiment shown, rotors 31, 32 are tiltable with respect to respective axes C integrally with respective nacelles 11, 12. Therefore, sensor means 41 are also configured to detect nacelle angles of nacelles 11, 12 corresponding to the orientation of nacelles 11, 12 with respect to axis A. Each engine 21, 22 is adapted to generate a respective engine torque TQ _eng1 , TQ _eng2 , which is transmitted at least in part to rotors 31, 32 by transmission units 6. In addition, each rotor 31, 32 is rotatable relative to the respective rotational axis B under the action of respective rotor torques TQ _mast1 , TQ _mast2 , which are correlated to engine torques TQ _eng1 and/or TQ _eng2 . In particular, in the event of an increase in engine torques TQ _eng1 , TQ _eng2 , collective pitch angles θ1 _COLL , θ2 _COLL are increased in such a way that angular speeds ω1, ω2 are kept constant due to the corresponding increase in the drag torques acting on rotors 31, 32. Furthermore, rotors 31, 32 are operatively connected to each other. In particular, convertiplane 1 comprises an interconnection shaft 4, which is operatively connected to rotors 31 and 32 (Figure 1). Interconnection shaft 4 is adapted to allow the rotation of rotors 31, 32 in case of failure of one of engines 21, 22. In a known manner, interconnection shaft 4 is adapted to: - allow at least part of engine torques TQ _eng1 , TQ _eng2 to be transmitted between transmission units 6 of rotors 31 and 32; and - allow rotors 31, 32 to tilt with respect to the respective axis C. Each rotor torque TQ _mast1 , TQ _mast2 is in general different from engine torques TQ _eng1 , TQ _eng2 for several reasons, which will be described in the following. In particular, at least part of engine torques TQ _eng1 , TQ _eng2 transmitted by transmission unit 6 is dissipated because of mechanical losses occurring at transmission unit 6. In addition, convertiplane 1 may comprise a plurality of not-shown accessories, which perform various functions on board the convertiplane and require a certain amount of power to be operated. The accessories are at least indirectly operatively connected to engines 21 and/or 22 to receive the power necessary for their operation. In particular, engines 21 and/or 22 are adapted to provide the accessories with a torque TQ _acc for powering them. Furthermore, convertiplane 1 comprises (Figure 4): - sensor means 42 configured to measure engine torques TQ _eng1 , TQ _eng2 generated by respective engines 21, 22; and - sensor means 43 configured to measure the airspeed v of convertiplane 1. Sensor means 42 and 43 are operatively connected to control unit 5 and are adapted to generate respective signals associated with the measured values of engine torques TQ _eng1 , TQ _eng2 and airspeed v of convertiplane 1. Control unit 5 is configured to calculate a first component TQ _mastSYM of rotor torques TQ _mast1 , TQ _mast2 on the basis of engine torques TQ _eng1 , TQ _eng2 . In detail, first component TQ _mastSYM is equal to rotor torques TQ _mast1 , TQ _mast2 when collective pitch angles θ1 _COLL , θ2 _COLL are equal to each other. For the purpose of calculating first component TQ _mastSYM , control unit 5 is configured to (Figure 2): - receive from sensor means 42 the signal associated with the measured values of engine torques TQ _eng1 , TQ _eng2 ; - calculating a total engine torque TQ _engTOT generated by engines 21 and 22 as the sum of the engine torques TQ _eng1 , TQ _eng2 (block 50); - subtracting from the total engine torque TQ _engTOT a term corresponding to the transmission losses due to the transmission of engine torques TQ _eng1 , TQ _eng2 (or part thereof) from engines 21, 22 to rotors 31, 32 (block 51) through transmission unit 6; and/or - subtracting from the total engine torque TQ _engTOT the torque TQ _acc for powering the accessories of convertiplane 1, where present (block 52). In particular, control unit 5 may be configured to subtract the term corresponding to the transmission losses from the total engine torque TQ _engTOT by multiplying total engine torque TQ _engTOT by a mechanical loss coefficient η _XMNS (η _XMNS < 1). The difference between total engine torque TQ _engTOT and the transmission losses and/or torque TQ _acc is the total available engine torque TQ _engavail . Control unit 5 is further configured to calculate first component TQ _mastSYM by dividing the total available engine torque TQ _engavail by two, which is the number of rotors 31, 32 of convertiplane 1 (block 53 in Figure 2). Advantageously, control unit 5 is configured to: - receive the signal associated with collective pitch angles θ1 _COLL , θ2 _COLL of rotors 31 and 32 from sensor means 40; and - calculate a second component TQ _mastASYM1 , TQ _mastASYM2 of respective rotor torques TQ _mast1 , TQ _mast2 on the basis of at least a respective pitch angle difference Δθ _COLL1, Δθ _COLL2 between collective pitch angles θ1 _COLL and θ2 _COLL (block 54 of Figure 2). In particular, second components TQ _mastASYM1 , TQ _mastASYM2 respectively represent how much the rotor torque TQ _mast1 , TQ _mast2 of each rotor 31, 32 deviates from the rotor torque that would act on each rotor 31, 32 if total available engine torque TQ _engavail were equally apportioned between rotors 31 and 32. Accordingly, first component TQ _mastSYM and second components TQ _mastASYM1 , TQ _mastASYM2 , may be referred respectively to as a “symmetric component” and an “asymmetric components” of rotor torques TQ _mast1 , TQ _mast2 . The following are exemplary situations in which convertiplane 1 is in a flight condition in which total available engine torque TQ _engavail is not equally apportioned between rotors 31 and 32: - hovering during vertical take-off and landing or during conversion of convertiplane 1 between the airplane and the helicopter configurations with steady cross wind; - nacelle angle equal to 75° with steady heading sideslip relative to a direction fixed with respect to the ground; - nacelle angle equal to 50° with steady heading sideslip relative to the direction fixed with respect to the ground; - convertiplane 1 arranged in the airplane configuration with steady heading sideslip relative to the direction fixed with respect to the ground; - bank angle different than zero and nacelle angle equal to 50° to maintain null sideslip. Control unit 5 is further configured to calculate each rotor torque TQ _mast1 , TQ _mast2 as the sum of first component TQ _mastSYM and the respective second component TQ _mastASYM1 , TQ _mastASYM2 (TQ _mast1 = TQ _mastSYM + TQ _mastASYM1; TQ _mast2 = TQ _mastSYM + TQ _mastASYM2 ) (block 55 in Figure 2). More specifically, for the purpose of calculating second components TQ _mastASYM1, TQ _mastASYM2 (block 54), control unit 5 is configured to (Figure 2): - determine parameters TQ’ ₁ , TQ’ ₂ , which are respectively associated with the variation of rotor torques TQ _mast1 , TQ _mast2 as a result of a variation of collective pitch angle θ1 _COLL , θ2 _COLL of the same rotor 31, 32 (block 56); and - multiply each parameter TQ’ ₁ , TQ’ ₂ by respective pitch angle difference Δθ _COLL1, Δθ _COLL2 (block 57). Pitch angle differences Δθ _COLL1 , Δθ _COLL2 of rotors 31, 32 are respective differences between each collective pitch angle θ1 _COLL , θ2 _COLL and a symmetric collective pitch angle θ0 (Δθ _COLL1 = θ1 _{COLL -} θ0; Δθ _COLL2 = θ2 _{COLL -} θ0). In detail, symmetric collective pitch angle θ0 is calculated as the sum of collective pitch angles θ1 _COLL , θ2 _COLL divided by two, which is the number of rotors 31, 32 of aircraft 1 (θ0 = (θ1 _COLL + θ2 _COLL )/2).In detail, each parameter TQ’ ₁ , TQ’ ₂ may be expressed as a partial derivative of the relative rotor torque TQ _mast1 , TQ _mast2 with respect to the collective pitch angle θCOLL (TQ’ ₁ = ∂TQ _mast1 /∂θ _COLL ; TQ’ ₂ = ∂TQ _mast2 /∂θ _COLL ). Parameters TQ’ ₁ , TQ’ ₂ are variable as a function of several variables associated with the flight conditions of convertiplane 1. In particular, the memory of control unit 5 comprises a database 9, in which data correlating parameters TQ’ ₁ , TQ’ ₂ with a plurality of variables associated with the flight conditions of convertiplane 1 are stored. In detail, the data stored in database 9 may be arranged in tables and/or graphs, which are preferably multiple input tables and/or graphs (Figure 4). Control unit 5 is configured to: - receive a plurality of signals associated with the flight conditions of convertiplane 1; - access database 9 using the signals associated with the flight conditions as inputs; and - determine parameters TQ’ ₁ , TQ’ ₂ on the basis of the inputs and on the basis of the data correlating parameters TQ’ ₁ , TQ’ ₂ and the variables associated with the flight conditions of convertiplane 1 stored in database 9. More specifically, the data stored in database 9 are calculated or experimentally measured in a known manner by means of statistical methods or by test on the basis of one or more sets of the following variables of convertiplane 1: - airspeed; - torque settings; - nacelle angle; - rotor speed; - altitude; - ambient temperature; - number of operative engines 21, 22. With reference to Figure 3, control unit 5 is configured to calculate parameters TQ’ ₁ , TQ’ ₂ on the basis of at least two independent variables associated with the performance of rotor 31 and/or 32. In the embodiment shown in Figure 3, each parameter TQ’ ₁ , TQ’ ₂ is calculated on the basis of: - the angle α1, α2 of the respective rotor 31, 32 measured by sensor means 41; and - airspeed v measured by sensor means 43; and/or - first component TQ _mastSYM . In detail, database 9 comprises data correlating parameters TQ’ ₁ , TQ’ ₂ with respective angle α1, α2, airspeed v and first component TQ _mastSYM . In further detail, control unit 5 is configured to (Figure 3): - calculate parameter TQ’ ₁ , TQ’ ₂ on the basis of airspeed v, if respective angle α1, α2 is lower than or equal to a threshold value th (block 58); and - calculate parameter TQ’ ₁ , TQ’ ₂ on the basis of first component TQ _mastSYM , if respective angle α1, α2 is greater than threshold value th (block 59). In particular, if the collective pitch angle θ1 _COLL , θ2 _COLL is the same for the two rotors 31, 32, the second components TQ _mastASYM1 , TQ _mastASYM2 are both null and the rotor torques TQ _mast1 , TQ _mast2 are equal to each other and to the first component TQ _mastSYM (TQ _mast1 = TQ _mast2 = TQ _mastSYM ). In addition, since rotors 31, 32 are identical to each other and have the same power required for the same input, if angles α1 and α2 are equal to each other, parameters TQ’ ₁ , TQ’ ₂ are also identical. In other words, if angles α1 and α2 are equal to each other, the variation of rotor torque TQ _mast1 as a result of a variation of collective pitch angle θ1 _COLL is identical to the variation of rotor torque TQ _mast2 as a result of a variation of collective pitch angle θ2 _COLL . In use, control unit 5 calculates an estimate of rotor torques TQ _mast1 , TQ _mast2 with a sum of first component TQ _mastSYM and respective second components TQ _mastASYM1 , TQ _mastASYM2 . The following is a description of the steps required to calculate first component TQ _mastSYM . In detail, sensor means 42 periodically measure engine torques TQ _eng1 , TQ _eng2 andsend the associated signal to control unit 5 (Figure 4). Subsequently, control unit 5 calculates total engine torque TQ _engTOT as the sum of the engine torques TQ _eng1 , TQ _eng2 measured by sensor means 42 (block 50 in Figure 2). Control unit 5 then calculates total available engine torque TQ _engavail by subtracting from total engine torque TQ _engTOT the transmission losses due to the transmission of engine torques TQ _eng1 , TQ _eng2 through transmission unit 6 (block 51 in Figure 2) and/or the torque TQ _acc for powering the accessories of convertiplane 1, where present (block 52 in Figure 2). Control unit 5 then calculates first component TQ _mastSYM by dividing available engine torque TQ _engavail by the number of rotors 31, 32 of convertiplane 1 (block 53 in Figure 2). The following is a description of the steps required to calculate second components TQ _mastASYM1 , TQ _mastASYM2 . Periodically, sensor means 40 measure collective pitch angles θ1 _COLL , θ2 _COLL of rotors 31 and 32, sensor means 41 measure angles α1, α2 and sensor means 43 measure airspeed v. Control unit 5 calculates the pitch angle differences Δθ _COLL1 , Δθ _COLL2 between collective pitch angles θ1 _COLL and θ2 _COLL . In addition, control unit 5 accesses database 9 using angle α1, α2, airspeed v and/or first component TQ _mastSYM as inputs and determines parameters TQ’ ₁ , TQ’ ₂ on the basis of the data stored in database 9. In particular, if angle α1 is lower than or equal to threshold value th, control unit 5 calculates parameter TQ’ ₁ on the basis of angle α1 and airspeed v (block 58 in Figure 3); if angle α1 is equal to or greater than threshold value th, control unit 5 calculates parameter TQ’ ₁ on the basis of angle α1 and first component TQ _mastSYM (block 59 in Figure 3). Similarly to the foregoing, if angle α2 is lower than or equal to threshold value th, control unit 5 calculates parameter TQ’ ₂ on the basis of angle α2 and airspeed v; if angle α2 is equal to or greater than threshold value th, control unit 5 calculates parameter TQ’ ₂ on the basis of angle α2 and first component TQ _mastSYM . Once parameters TQ’ ₁ , TQ’ ₂ are determined, control unit 5 calculates each second component TQ _mastASYM1 , TQ _mastASYM2 by multiplying respective parameter TQ’ ₁ , TQ’ ₂ by the respective pitch angle difference Δθ _COLL1 , Δθ _COLL2 (block 57 in Figure 2). Control unit 5 then calculates each rotor torque TQ _mast1 , TQ _mast2 as the sum of first component TQ _mastSYM and the respective second component TQ _mastASYM1 , TQ _mastASYM2 (block 55 in Figure 2). With reference to Figure 5, 1’ denotes an aircraft capable of hovering according to a second embodiment of the present invention. Aircraft 1’ is similar to aircraft 1 and will be described hereinafter only insofar as it differs from the latter; equal or equivalent parts of aircrafts 1, 1’ will be marked, where possible, by the same reference numerals. Aircraft 1’ differs from aircraft 1 in that it comprises: - two pairs of half-wings 3, 3’; the half-wings of each pair of half-wings 3, 3’ extending cantilevered from respective parts opposite one another of fuselage 2 and transversely to axis A; the two pairs of half-wings 3, 3’ being distanced from each other along axis A; - four rotors 31, 32, 33’, 34’ operatively connected with respective engines 21, 22, 23’, 24’ (only schematically shown in figure 7). Each rotor 31, 32, 33’, 34’ essentially comprises a hub 7 and a plurality of blades 8. In addition, rotors 31, 32, 33’, 34’ are rotatable about respective axes B with respective angular speeds ω1, ω2, ω3 and ω4, which are preferably constant over time. Furthermore, blades 8 are articulated on the respective hub 7 in such a way that respective angles of attack θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL are collectively adjustable relative to respective axes B. Rotors 31, 32, 33’, 34’ are controllable independently of one another. In particular, the collective pitch angles θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL may be different from one another. Furthermore, rotors 31, 32, 33’, 34’ are identical to one another. In detail, rotors 31, 32, 33’, 34’ have the same power required for the same input. In addition, rotors 31, 32 are tiltable with respect to respective axes C (not-shown) relative to half-wings 3 and rotors 33’, 34’ are tiltable with respect to respective axes C (not-shown) relative to half-wings 3’. In detail, aircraft 1’ is a quadcopter. Aircraft 1’ further comprises (Figure 7): - sensor means 40 configured to measure collective pitch angles θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL ; - sensor means 41 configured to measure angles α1, α2, α3, α4 of respective rotational axes B of rotors 31, 32, 33’, 34’ with respect to axis A. Each engine 21, 22, 23’, 24’ is adapted to generate a respective engine torque TQ _eng1 , TQ _eng2 , TQ _eng3 , TQ _eng4 , which is transmitted at least in part to rotors 31, 32, 33’, 34’ by respective transmission units 6. In addition, each rotor 31, 32, 33’, 34’ is rotatable relative to the respective rotational axis B under the action of respective rotor torques TQ _mast1 , TQ _mast2 , TQ _mast3 , TQ _mast4 , which are correlated to engine torques TQ _eng1 and/or TQ _eng2 and/or TQ _eng3 and/or TQ _eng4 . Furthermore, rotors 31, 32, 33’, 34’ are operatively connected to one another. In particular, convertiplane 1’ comprises an interconnection mechanism 4’, which is operatively connected to rotors 31, 32, 33’, 34’ (Figure 5). Interconnection mechanism 4’ is adapted to allow the rotation of rotors 31, 32, 33’, 34’ in case of failure of one of engines 21, 22, 23’, 24’. In detail, each engine 21, 22, 23’, 24’ is coaxial to the respective rotor 31, 32, 33’, 34’ and interconnection mechanism 4’ comprises (Figure 5): - an interconnection shaft 4a’, which operatively connects engine 21 to engine 22; - an interconnection shaft 4b’, which operatively connects engine 23’ to engine 24’; and - an interconnection shaft 4c’, which operatively connects interconnection shaft 4a’ to interconnection shaft 4b’. Aircraft 1’ further comprises (Figure 7): - sensor means 42 configured to measure engine torques TQ _eng1 , TQ _eng2 , TQ _eng3 , TQ _eng4 generated by respective engines 21, 22, 23’, 24’; and - sensor means 43 configured to measure the airspeed v of aircraft 1’. Control unit 5 is configured to calculate a first component TQ _mastSYM of rotor torques TQ _mast1 , TQ _mast2 , TQ _mast3 , TQ _mast4 on the basis of engine torques TQ _eng1 , TQ _eng2 , TQ _eng3 , TQ _eng4 . In detail, first component TQ _mastSYM is equal to rotor torques TQ _mast1 , TQ _mast2 , TQ _mast3 , TQ _mast4 when collective pitch angles θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL are equal to one another. For the purpose of calculating first component TQ _mastSYM , control unit 5 is configured to (Figure 6): - receive from sensor means 42 the signal associated with the measured values of engine torques TQ _eng1 , TQ _eng2 , TQ _eng3 , TQ _eng4 ; - calculating a total engine torque TQ _engTOT generated by engines 21, 22, 23’ and 24’ as the sum of the engine torques TQ _eng1 , TQ _eng2 , TQ _eng3 , TQ _eng4 (block 50); - subtracting from the total engine torque TQ _engTOT a term corresponding to the transmission losses due to the transmission of engine torques TQ _eng1 , TQ _eng2, TQ _eng3 , TQ _eng4 (or part thereof) from engines 21, 22, 23’, 24’ to rotors 31, 32, 33’, 34’ (block 51) through transmission units 6; and/or - subtracting from the total engine torque TQ _engTOT the torque TQ _acc for powering the accessories of aircraft 1’, where present (block 52). The difference between total engine torque TQ _engTOT and the transmission losses and/or torque TQ _acc is the total available engine torque TQ _engavail . Control unit 5 is further configured to calculate first component TQ _mastSYM by dividing the total available engine torque TQ _engavail by four, which is the number of rotors 31, 32, 33’, 34’ of aircraft 1’ (block 53 in Figure 6). Advantageously, control unit 5 is configured to: - receive the signal associated with collective pitch angles θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL of rotors 31, 32, 33’, 34’ from sensor means 40; and - calculate a second component TQ _mastASYM1 , TQ _mastASYM2 , TQ _mastASYM3 , TQ _mastASYM4 of each rotor torque TQ _mast1 , TQ _mast2 , TQ _mast3 , TQ _mast4 on the basis of at least pitch angle differences Δθ _COLL1 , Δθ _COLL2 , Δθ _COLL3 , Δθ _COLL4 between collective pitch angles θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL (block 54 of Figure 6). In particular, second components TQ _mastASYM1 , TQ _mastASYM2 , TQ _mastASYM3 , TQ _mastASYM4 represent how much the rotor torque TQ _mast1 , TQ _mast2 , TQ _mast3 , TQ _mast4 of each rotor 31, 32, 33’, 34’ deviates from the rotor torque that would act on each rotor 31, 32, 33’, 34’ if total available engine torque TQ _engavail were equally apportioned between rotors 31, 32, 33’, 34’. Control unit 5 is further configured to calculate rotor torques TQ _mast1 , TQ _mast2 , TQ _mast3 , TQ _mast4 as an algebraic sum of first component TQ _mastSYM (which is common to all rotors 31, 32, 33’, 34’) and the respective second component TQ _mastASYM1 , TQ _mastASYM2 , TQ _mastASYM3 , TQ _mastASYM4 (TQ _mast1 = TQ _mastSYM + TQ _mastASYM1 ; TQ _mast2 = TQ _mastSYM + TQ _mastASYM2 ; TQ _mast3 = TQ _mastSYM + TQ _mastASYM3 ; TQ _mast4 = TQ _mastSYM + TQ _mastASYM4 ;) (block 55 in Figure 6). In particular, if the collective pitch angle θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL is the same for the four rotors, the second components TQ _mastASYM1 , TQ _mastASYM2 , TQ _mastASYM3 , TQ _mastASYM4 are null and the rotor torques TQ _mast1 , TQ _mast2 , TQ _mast3 , TQ _mast4 are equal to one another and to the first component TQ _mastSYM (TQ _mast1 = TQ _mast2 = TQ _mast3 = TQ _mast4 = TQ _mastSYM ). More specifically, for the purpose of calculating the second component TQ _mastASYM1 , TQ _mastASYM2 , TQ _mastASYM3 , TQ _mastASYM4 of each rotor 31, 32, 33’, 34’ (block 54), control unit 5 is configured to (Figure 6): - determine parameters TQ’ ₁ , TQ’ ₂ , TQ’ ₃ , TQ’ ₄ , which are respectively associated with the variation of the respective rotor torque TQ _mast1 , TQ _mast2 , TQ _mast3 , TQ _mast4 as a result of a variation of collective pitch angle θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL of the same rotor 31, 32, 33’, 34’ (block 56); and - multiply each parameter TQ’ ₁ , TQ’ ₂ , TQ’ ₃ , TQ’ ₄ by the pitch angle difference Δθ _COLL1 , Δθ _COLL2 , Δθ _COLL3 , Δθ _COLL4 of the respective rotor 31, 32, 33’, 34 (block 57). Control unit 5 is configured to calculate each parameter TQ’ ₁ , TQ’ ₂ , TQ’ ₃ , TQ’ ₄ on the basis of at least two independent variables associated with the performance of the respective rotor 31, 32, 33’, 34’. In detail, each parameter TQ’ ₁ , TQ’ ₂ , TQ’ ₃ , TQ’ ₄ is calculated on the basis of: - the respective angle α1, α2, α3, α4 measured by sensor means 41; and - airspeed v measured by sensor means 43; and/or - first component TQ _mastSYM . In detail, database 9 comprises data correlating parameter TQ’ with angle α1, α2, α3, α4, airspeed v and first component TQ _mastSYM . In further detail, control unit 5 is configured to: - calculate each parameter TQ’ ₁ , TQ’ ₂ , TQ’ ₃ , TQ’ ₄ on the basis of airspeed v, if the respective angle α1, α2, α3, α4 is lower than or equal to a threshold value th; and - calculate each parameter TQ’ ₁ , TQ’ ₂ , TQ’ ₃ , TQ’ ₄ on the basis of first component TQ _mastSYM , if the respective angle α1, α2, α3, α4 is greater than threshold value th. Since rotors 31, 32, 33’, 34’ are identical to one another, if angles α1, α2, α3, α4 are equal to one another, parameters TQ’ ₁ , TQ’ ₂ , TQ’ ₃ , TQ’ ₄ are also equal to one another. Pitch angle differences Δθ _COLL1 , Δθ _COLL2 , Δθ _COLL3 , Δθ _COLL4 of rotors 31, 32, 33’, 34’ are respective differences between each collective pitch angle θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL and a symmetric collective pitch angle θ0 (Δθ _COLL1 = θ1 _{COLL -} θ0; Δθ _COLL2 = θ2 _{COLL -} θ0; Δθ _COLL3 = θ3 _{COLL -} θ0; Δθ _COLL4 = θ4 _{COLL -} θ0). In detail, symmetric collective pitch angle θ0 is calculated as the sum of collective pitch angles θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL divided by four, which is the number of rotors 31, 32, 33’, 34’ of aircraft 1’ (θ0 = (θ1 _COLL + θ2 _COLL + θ3 _COLL + θ4 _COLL )/4). In use, control unit 5 calculates an estimate of rotor torques TQ _mast1 , TQ _mast2 , TQ _mast3 , TQ _mast4 as an algebraic sum of first component TQ _mastSYM and respective second components TQ _mastASYM1 , TQ _mastASYM2 , TQ _mastASYM3 , TQ _mastASYM4 . The description of the steps required to calculate first component TQ _mastSYM of rotor torques TQ _mast1 , TQ _mast2 , TQ _mast3 , TQ _mast4 of aircraft 1’ is similar to the description of the steps required to calculate first component TQ _mastSYM of rotor torques TQ _mast1 , TQ _mast2 of aircraft 1 and will be omitted for the sake of brevity. The following is a description of the steps required to calculate second components TQ _mastASYM1 , TQ _mastASYM2 , TQ _mastASYM3 , TQ _mastASYM4 . Periodically, sensor means 40 measure collective pitch angles θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL of rotors 31, 32, 33’, 34’, sensor means 41 measure angles α1, α2, α3, α4 and sensor means 43 measure airspeed v. Control unit 5 calculates pitch angle differences Δθ _COLL1 , Δθ _COLL2 , Δθ _COLL3 , Δθ _COLL4 between collective pitch angles θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL . In addition, control unit 5 accesses database 9 using angle α1, α2, α3, α4, airspeed v and/or first component TQ _mastSYM as inputs and determines parameters TQ’ ₁ , TQ’ ₂ , TQ’ ₃ , TQ’ ₄ on the basis of the data stored in database 9. In detail, each pitch angle difference Δθ _COLL1 , Δθ _COLL2 , Δθ _COLL3 , Δθ _COLL4 is calculated by subtracting symmetric collective pitch angle θ0 from collective pitch angles θ1 _COLL , θ2 _COLL , θ3 _COLL , θ4 _COLL of respective rotors 31, 32, 33’, 34’. Subsequently, control unit 5 calculates second components TQ _mastASYM1 , TQ _mastASYM2 , TQ _mastASYM3 , TQ _mastASYM4 as multiplications of respective parameters TQ’ ₁ , TQ’ ₂ , TQ’ ₃ , TQ’ ₄ by respective pitch angle differences Δθ _COLL1 , Δθ _COLL2 , Δθ _COLL3 , Δθ _COLL4 . Control unit 5 then calculates each rotor torque TQ _mast1 , TQ _mast2 , TQ _mast3 , TQ _mast4 as the sum of first component TQ _mastSYM and the respective second component TQ _mastASYM1 , TQ _mastASYM2 , TQ _mastASYM3 , TQ _mastASYM4 . Figure 8 illustrates a block diagram related to a method for estimating the rotor torques of an aircraft 1’’ capable of hovering (not shown) according to a second embodiment of the present invention comprising N operatively interconnected and identical rotors. In detail, N is a natural number greater than one. In addition, N may be an even or an odd number. Aircraft 1’’ further comprises M engines operatively connected to the N rotors. In detail, M is a natural number greater than or equal to one; M might be equal to or different from N. Aircraft 1’’ is a generalization of aircrafts 1 and 1’. Accordingly, equal or equivalent parts of aircrafts 1, 1’, 1’’ will be marked, where possible, by the same reference numerals. In detail, aircraft 1’’ is a multicopter. According to the estimation method shown in Figure 8, control unit 5 of aircraft 1’’ is configured to calculate a first component TQ _mastSYM of rotor torques TQ _mast1 , TQ _mast2 ,…, TQ _mastN on the basis of engine torques TQ _eng1 , TQ _eng2 ,…, TQ _engM . In detail, first component TQ _mastSYM is equal to rotor torques TQ _mast1 , TQ _mast2 ,…, TQ _mastN when collective pitch angles θ1 _COLL , θ2 _COLL ,…, θN _COLL are equal to one another. For the purpose of calculating first component TQ _mastSYM , control unit 5 is configured to (Figure 8): - receive from sensor means 42 the signal associated with the measured values of engine torques TQ _eng1 , TQ _eng2 ,…, TQ _engM ; - calculating a total engine torque TQ _engTOT generated by the engines operatively connected to the rotors of aircraft 1’’ as the sum of the engine torques TQ _eng1 , TQ _eng2 ,…,TQ _engM (block 50); - subtracting from the total engine torque TQ _engTOT a term corresponding to the transmission losses due to the transmission of engine torques TQ _eng1 , TQ _eng2, …, TQ _engM (or part thereof) from the M engines (block 51) through transmission units 6; and/or - subtracting from the total engine torque TQ _engTOT the torque TQ _acc for powering the accessories of aircraft 1’’, where present (block 52). The difference between total engine torque TQ _engTOT and the transmission losses and/or torque TQ _acc is the total available engine torque TQ _engavail . Control unit 5 is further configured to calculate first component TQ _mastSYM by dividing the total available engine torque TQ _engavail by the number of rotors N of aircraft 1’’ (block 53 in Figure 8, formula a)): Advantageously, control unit 5 is configured to: - receive the signal associated with collective pitch angles θ1 _COLL , θ2 _COLL ,.., θN _COLL of the N rotors from sensor means 40; and - calculate a second component TQ _mastASYM1 , TQ _mastASYM2 ,…, TQ _mastASYMN of each rotor torque TQ _mast1 , TQ _mast2 ,…, TQ _mastN on the basis of at least pitch angle differences Δθ _COLL1 , Δθ _COLL2 ,…, Δθ _COLLN between collective pitch angles θ1 _COLL , θ2 _COLL ,…, θN _COLL (block 54 of Figure 8). In particular, second components TQ _mastASYM1 , TQ _mastASYM2 ,…, TQ _mastASYMN represent how much the rotor torque TQ _mast1 , TQ _mast2 ,…, TQ _mastN of each rotor deviates from the rotor torque that would act on each of the rotors if total available engine torque TQ _engavail were equally apportioned between the rotors. Control unit 5 is further configured to calculate rotor torques TQ _mast1 , TQ _mast2 ,…, TQ _mastN as an algebraic sum of first component TQ _mastSYM (which is common to all rotors of aircraft 1’’) and the respective second component TQ _mastASYM1 , TQ _mastASYM2 ,…, TQ _mastASYMN (block 55 in Figure 8, formula b)): In particular, if the collective pitch angle θ1 _COLL , θ2 _COLL ,…, θN _COLL is the same for the N rotors, the second components TQ _mastASYM1 , TQ _mastASYM2 ,…, TQ _mastASYMN are null and the rotor torques TQ _mast1 , TQ _mast2 ,…, TQ _mastN are equal to one another and to the first component TQ _mastSYM (TQ _mast1 = TQ _mast2 = … = TQ _mastN = TQ _mastSYM ). More specifically, for the purpose of calculating second components TQ _mastASYM1 , TQ _mastASYM2 ,…, TQ _mastASYMN , control unit 5 is configured to (Figure 8): - determine parameters TQ’ ₁ , TQ’ ₂ ,…,TQ’ _N , which are respectively associated with the variation of the respective rotor torque TQ _mast1 , TQ _mast2 ,…, TQ _mastN as a result of a variation of collective pitch angle θ1 _COLL , θ2 _COLL ,…, θN _COLL of the same rotor (block 56); and - multiply each parameter TQ’ ₁ , TQ’ ₂ ,…,TQ’ _N by the respective pitch angle difference Δθ _COLL1 , Δθ _COLL2 ,…, Δθ _COLLN (block 57, formula c)): Control unit 5 is configured to calculate each parameter TQ’ ₁ , TQ’ ₂ ,…, TQ’ _N on the basis of at least two independent variables associated with the performance of the respective rotor. In detail, each parameter TQ’ ₁ , TQ’ ₂ ,…,TQ’ _N is calculated on the basis of: - the angle α1, α2,…, αN of the respective rotor measured by sensor means 41; and - airspeed v measured by sensor means 43; and/or - first component TQ _mastSYM . In detail, database 9 comprises data correlating parameters TQ’ ₁ , TQ’ ₂ ,…,TQ’ _N with angles α1, α2,…,αN, airspeed v and first component TQ _mastSYM . In further detail, control unit 5 is configured to: - calculate each parameter TQ’ ₁ , TQ’ ₂ ,…, TQ’ _N on the basis of airspeed v, if the respective angle α1, α2,…, αN is lower than or equal to a threshold value th; and - calculate each parameter TQ’ ₁ , TQ’ ₂ ,…, TQ’ _N on the basis of first component TQ _mastSYM , if the respective angle α1, α2,…, αN is greater than threshold value th. Since rotors N are identical to one another, if angles α1, α2,…, αN are equal to one another, parameters TQ’ ₁ , TQ’ ₂ ,…, TQ’ _N are also equal to one another. Pitch angle differences Δθ _COLL1 , Δθ _COLL2 ,…, Δθ _COLLN are respective differences between each collective pitch angle θ1 _COLL , θ2 _COLL ,…, θ4 _COLL and a symmetric collective pitch angle θ0 (formula d)): In detail, symmetric collective pitch angle θ0 is calculated as the sum of collective pitch angles θ1 _COLL , θ2 _COLL ,…, θN _COLL divided by the number N of rotors (formula e)): From an examination of the characteristics of the method for estimating the rotor torques TQ _mast1 , TQ _mast2 ,..., TQ _mastN, control unit 5 and aircraft 1, 1’, 1’’ according to the present invention, the advantages they allow obtaining are evident. In particular, rotor torques TQ _mast1 , TQ _mast2 of convertiplane 1 are calculated as the algebraic sum of first component TQ _mastSYM , which is equal to rotor torques TQ _mast1 , TQ _mast2 when collective pitch angles θ1 _COLL , θ2 _COLL are equal to each other, and respective second components TQ _mastASYM1 , TQ _mastASYM2 , which represents how much the rotor torque TQ _mast1 , TQ _mast2 of each rotor 31, 32 deviates from the rotor torque that would act on each rotor 31, 32 if total available engine torque TQ _engavail were equally apportioned between rotors 31 and 32. Therefore, the rotor torques TQ _mast1 , TQ _mast2 of rotors 31, 32, which are independently controllable from each other, can be efficiently and reliably estimated even in situations in which collective pitch angles θ1 _COLL , θ2 _COLL are different from each other and without any physical sensor for directly measuring the rotor torques TQ _mast1 , TQ _mast2 . Indeed, it has been observed that the rotor torques TQ _mast1 , TQ _mast2 estimated by combining first and second components TQ _mastSYM , TQ _mastASYM1, TQ _mastASYM2 are comparable to and more reliable than the rotor torques directly measured by the physical sensors mentioned in the introductory part of the description. The estimation method according to the present invention also allows the rotor torques of an aircraft comprising any number of rotors to be efficiently estimated. Furthermore, control unit 5 comprises database 9, in which data correlating parameters TQ’ ₁ , TQ’ ₂ ,…,TQ’ _N with variables associated with the flight conditions of convertiplane 1 are stored. Therefore, parameters TQ’ ₁ , TQ’ ₂ ,…,TQ’ _N and second components TQ _mastASYM1 , TQ _mastASYM2 ,…, TQ _mastASYMN are calculable on the basis of a limited number of independent variables. The fact that parameters TQ’ ₁ , TQ’ ₂ ,…,TQ’ _N can be determined on the basis of a limited number of variables allows to improve the robustness of the estimation method, because fewer errors due to false dependencies on other variables are possible. In addition, since angles α1, α2,..., αN corresponding to the orientation of axes B of the respective rotors are taken into account when determining parameters TQ’ ₁ , TQ’ ₂ ,…,TQ’ _N , second components TQ _mastASYM1 , TQ _mastASYM2 ,…, TQ _mastASYMN are calculated regardless of whether the M engines are tilted integrally with the N rotors or are fixed with respect to fuselage 2. Finally, it is clear that modifications and variations can be made to the method, the control unit 5 and the aircraft 1, 1’, 1’’ previously described without thereby departing from the scope of protection of the present invention. In particular, the method according to the present invention may be implemented for estimating the rotor torques of a helicopter comprising at least two rotors, which are operatively connected to one another and have respective collective pitch angles independently controllable from one another. Furthermore, the method according to the present invention may be implemented for estimating the rotor torques of an aircraft capable of hovering comprising at least two operatively connected and independently controllable rotors and with hybrid propulsion or full electric propulsion. In detail, the aircraft may comprise one engine or more than one engines and/or one or more electric motors. Finally, nacelles 11, 12 and respective engines 21, 22 might be rotationally fixed with respect to axis B, being only rotors 31, 32 tiltable around axis B.