The end-effector has two rotational and one translational properly decoupled freedoms. It is suitable to carry and orient a rigid body in space as well as to operate a vision system for stereoscopic vision, particularly in underwater environments and/or environments characterised by high fluid dynamic resistance. This invention is also directed to a scanning methodology for stereoscopic vision; such scanning methodology is particularly suitable to the mechanics of said support device that is the object of the present invention.

Finally, this invention is directed to a 3-DoF (degree-of- freedom) spherical orienting device, which is a sub-part of said support device that is the object of the present invention.

BACKGROUND OF THE ART Suppose that the state of illumination of an environment cannot be changed, i. e. lights cannot be reoriented in order to modify the cast shadows. This is usual in wide environments and in environments filled by very light-absorbing media, such as water. Stereoscopic vision in such environments is possible if: more than one point of view is available, in order to compare the images of same objects taken from different perspectives and compute their distances from the observer by triangulation; or if one point of view is available, provided that such point of view can be moved along the view axis. In the first case, the three-dimensional (3D) location of an object in the field of vision is deduced comparing the directions of at least two view axes converging on such object. This requires the use of at least two independent and properly controlled vision devices. In harsh environments and/or in demand of high reliability and control simplicity, stereoscopic vision by more viewpoints is unsuitable. In such cases it is preferable to use one vision device (one point of view) and localize the objects by comparing images where such objects appear with different geometric scale. Such pictures with differently scaled objects can be taken by rhythmical movement of the viewpoint along the view axis. So doing, the scale of an object changes as much as closer is such object to the viewpoint; hence, this allows computing the relative three-dimensional (3D) location of the objects in the field of vision.

The horizon of the viewpoint should be constant, as it is natural for a proper perception of the environment. Hence, the viewpoint should be oriented so that the rotation component along the view axis is always zero. This can be obtained in two ways : by a three rotational degree-of-freedom device, suitably controlled in order that the component of the resulting rotation of the viewpoint around the view axis is zero; by a two rotational degree-of-freedom device intrinsically lacking (due to its kinematics) of the rotational freedom about the view axis. In harsh environments, this second way is preferable, since it requires mechanisms with simpler architecture (two degree-of-freedom instead of three) and control is simpler.

Provided that a two rotational degree-of-freedom device is used, it is preferable that such two rotational freedoms are decoupled with respect to the horizon of the viewpoint for every direction of the view axis. Moreover, to simplify the scanning of the field of vision, it is preferable that the translational degree-of-freedom, which is required (as said before) for stereoscopic vision with one viewpoint, is decoupled from at least one of the two rotational freedoms of the device, preferably the rotation about the direction parallel to the horizon and orthogonal to the view axis (decoupled rotational freedom) ; i. e. that in every configuration one of the actuators commands said rotation about the direction parallel to the horizon and orthogonal to the view axis, while the two other actuators command both the rotation orthogonal to the horizon and orthogonal to the view axis, and the translation along the view axis. It is preferable that the decoupled rotational freedom is the rotational freedom parallel to the horizon of the viewpoint and orthogonal to the view axis, in order to proceed by horizontal scanning of the field of vision.

It is an object of the present invention to provide an Actuation System to orient and translate a rigid body in space, suitable for stereoscopic vision with one viewpoint and having all said characteristics for effective application in harsh environments. Such Actuation System is three degree-of- freedom: two rotational freedoms about two axes intersecting in a spherical centre of rotation fixed to the base; one translational freedom along a straight line through such spherical centre of rotation. These three freedoms are suitably decoupled, i. e. one of the two rotational freedoms is commanded independently to the other rotational freedom and to the translational freedom. Moreover, due to the kinematics of such Actuation System, rotation about the direction of the translational freedom is impossible in every configuration.

To be suitable for use in media characterised by high fluid dynamic resistance, mechanisms should be compact and their links should have small and properly shaped resistant surfaces. The parallel-hybrid architecture of the Actuation System, object of the present invention, has a high transmission gain (from actuators to end-effector); moreover, the structural stress of the links due to payload and fluid dynamic resistance on the submersed moving surfaces is low.

Hence it is possible to adopt slim links with armillary geometry in order to reduce the fluid dynamic resistance, so allowing higher velocities and accelerations of the end- effector. The actuators are fixed to the base, so that they do not increase the inertia of the moved links and can be easily hosted and protected inside a fluid-proof box.

DISCLOSURE OF THE INVENTION It is an object of the present invention to provide an Actuation System to orient and translate a rigid body in space.

Such Actuation System comprises: a base; first, second and third rotary actuators (the first, second and third rotary actuators of the Actuation System), each actuator having a shaft disposed in order to rotate about an actuation axis of rotation, each actuator being fixed to the base with actuator axes intersecting at a spherical centre of rotation (the spherical centre of the Actuation System) ; two closed, 2-DoF (degree-of- freedom), spherical kinematic chains (the spherical chains of the Actuation System), both with the same architecture and closed through the base, each composed of five links, the fifth link being common to both said spherical chains of the Actuation System and being connected at one side to the shaft of said third rotary actuator of the Actuation System, the first link of one said spherical chain of the Actuation System being connected at one side to the shaft of said first rotary actuator of the Actuation System while the first link of the other spherical chain of the Actuation System is connected at one side to the shaft of said second rotary actuator of the Actuation System, in each said spherical chain of the Actuation System the first link being connected to the second, the second to the third, the third to the fourth, the fourth to the fifth, by revolute joints whose axes pass all through said spherical centre of rotation of the Actuation System, in each said spherical chain of the Actuation System the axis of the revolute joint that connects the second link to the third being coincident to the axis of the revolute joint connecting the third link to the fourth (the output axis of the spherical chain), end-effector of each said spherical chain of the Actuation System being its third link, whose 2-DoF (degree-of-freedom) mobility comprises two rotational freedoms around said spherical centre of rotation of the Actuation System ; a 1-DoF (degree-of-freedom) mechanism (the translational section of the Actuation System) comprising the third links of both said spherical chains of the Actuation System, input coordinate of said translational section of the Actuation System being the angle that the output axis of one said spherical chain of the Actuation System forms with the output axis of the other spherical chain of the Actuation System, end-effector mobility of said translational section of the Actuation System being the translation along an axis (the orientation axis of the Actuation System), wherein said orientation axis of the Actuation System passes through said spherical centre of rotation of the Actuation System. The end-effector of said translational section of the Actuation System is also the end-effector of the Actuation System.

The two said spherical chains of the Actuation System share their common fifth link; together with the base, they constitute the spherical section of the Actuation System, which is a 3-DoF (degree-of-freedom) mechanism commanded by said three rotary actuators of the Actuation System. Output mobility of said spherical section of the Actuation System is the coupled orientation of both the third links of said spherical chains of the Actuation System, i. e. the coupled orientation of both said output axes of the spherical chains around said spherical centre of the Actuation System."Coupled orientation"means that it is not possible to command the orientation of said two axes of the spherical chains separately each other.

Provided that in both said spherical chains of the Actuation System the axes of the revolute joints that connect the third link to the fourth and the fourth link to the fifth are orthogonal, and provided that the axes of the joints that in both said spherical chains of the Actuation System connect the fourth link to the fifth link coincide, then said two output axes of the spherical chains and the axis of said third rotary actuator of the Actuation System lie, in a plane (the horizon plane of the Actuation System), whose orientation with respect to said base is commanded only by said third rotary actuator of the Actuation System. Nothing changes if in one said spherical chain of the Actuation System said third link is connected to the second link or the fourth link by a rigid connection in place of a revolute joint, although this can have an influence on the choice of said translational section of the Actuation System.

Obviously, in case that said spherical section of the Actuation System is used alone (without being connected to a translational section) and in one said spherical chain of the Actuation System said third link is connected by revolute joints to both said second and third links, then there is one passive mobility, which is in fact without influence.

It is a further object of the present invention to provide a particular embodiment of said Actuation System, the Armilleye, whose architecture is particularly suitable to carry and orient a camera as well as any other vision system specially for stereoscopic vision with one view point in environments characterised by high fluid-dynamic resistance and whose state of illumination cannot be changed. Armilleye uses as said translational section of the Actuation System a 1-DoF (degree of freedom), planar mechanism composed of two closed, interconnected chains lying in said horizon plane of the Actuation System that is the plane defined by the axis of said third rotary actuator of the Actuation System and by said orientation axis of the Actuation System, each such chain being composed of five links, two links (the fourth and fifth of each chain) being common to both chains, the first link being the third link of one said spherical chain of the spherical section of the Actuation System, the second link being connected to the first link and to the fifth link by revolute joints whose axes are parallel and orthogonal to said horizon plane of the Actuation System, the third link being connected to first and fourth links by revolute joints whose axis are parallel themselves and parallel to the axes of the revolute joints that connect the second link to first and fifth links, the fourth link being connected to the fifth by a cylindrical joint whose axis coincide with said orientation axis of the Actuation System or by a prismatic joint in the direction of said orientation axis of the Actuation System. The fifth link common to both said chains is the end-effector of the translational mechanism and is also the end-effector of Armilleye.

The vision system is carried by the end-effector. The view axis must coincide with the orientation axis of the Armilleye. So, the three actuators command the orientation of the vision system around said spherical centre of the Actuation System, without rotation component about said orientation axis of the Actuation System, and the horizon of the vision system coincide with the horizon of the Actuation System.

Since several applications require compact and simple vision systems, often small cameras, without embedded focus facilities, Armilleye can also be used to orient and support such vision systems. In the case of such application, the translational freedom along said orientation axis of the Armilleye is used to focus the camera. Analogously, said translational freedom of the Armilleye can be used as a zoom facility.

Instead of a vision system, said end-effector of the Actuation System can also carry and orient a spindle for machining as well as a gripper or a wrist in order to operate on the environment, e. g. for assembly or manipulation, or any other equipment requiring input spin and/or torque as well as a the third rotary degree of freedom around said spherical centre of rotation of the Actuation System. In the case that said end- effector carries a spindle, it is preferable that the rotation axis of such spindle coincides with said orientation axis of the Actuation System. Spin and/or torque are provided by a motor fixed to said base of the Actuation System and connected to said spindle through a cardan joint and an extensible limb. In order to not obstruct the mobility of the system, the centre of the cardan joint must coincide with said spherical centre of rotation of the Actuation System. In place of a cardan joint, it is also possible to use every other joint, provided that it can transmit torque, that it allows angular mobility about a fixed point and that such point coincides with said spherical centre of rotation of the Actuation System.

There are several advantages resulting from the design of said Actuation System and from the choice of the particular mechanism that Armilleye uses as said transational section of the Actuation System. Due to the architecture of said Actuation System, the degree-of-freedom of the end-effector makes easy the horizontal scanning of the field of vision, while shape and geometry of the links make such mechanism suitable for application in media with high fluid dynamic resistance.

Moreover, the particular mechanism that Armilleye uses as said transational section of the Actuation System is particularly suitable to host and carry a camera with high rigidity and satisfactory stroke.

It is a further object of the present invention to provide a scanning methodology of the field of vision of said Actuation System, and particularly of Armilleye, for stereoscopic vision of a large environment. Such scanning methodology consists of horizontal scanning along the horizon plane of the vision system, while the inclination of such horizon plane with respect to the horizon of the ground (horizon of the ground refers to a horizontal plane in the environment where the vision system operates) is changed by steps. Let us suppose that the environment to be scanned is larger than the field of vision of the vision system, so that it is not possible to shoot the entire environment at a time. In such case it is necessary to shoot sequences of images at different orientations of the vision system; such sequences of images are taken while the orientation of the view axis has been fixed and the vision system translates along such view axis. Each sequence of images allows reconstructing the three-dimensional (3D) location of the objects in the corresponding portion of the environment. So the three-dimensional location of the objects in the whole environment is reconstructed like a patchwork by putting together the data taken at different orientations of the view axis. The proposed scanning methodology consists of fixing the inclination of said horizon plane of the Actuation System with respect to the horizon of the ground by operating said third rotary actuator of the Actuation System, which commands the decoupled rotational freedom of said orientation axis of the Actuation System. Then the scanning proceeds in said horizon plane of the Actuation System by suitably operating said first and second rotary actuators of the Actuation System in order to: change the direction of said orientation axis of the Actuation System by angular steps ; and, at each step, translate the vision system along said orientation axis of the Actuation System to shoot a sequence of images.

After having completed this horizontal scanning of the environment, the inclination of said horizon plane of the Actuation System is changed of an angular step by operating said third rotary actuator of the Actuation System and a new horizontal scanning is performed. The span of the angular steps is chosen in order that a new portion of environment is enclosed in the field of vision of the vision system, with as small superimposition as possible.

Further details and advantages of the present invention will be apparent from the detailed description and drawings below.

BRIEF DESCRIPTION OF THE DRAWINGS In order that the invention may be readily understood, three preferred embodiments of the invention will be described by way of example, with reference to the accompanying drawings, wherein: FIG. 1 is an isometric view of a realistic embodiment of the invention with a particular said translational section of the Actuation System (Armilleye) adapted to orient a cylindrical camera payload for stereoscopic vision with one view point; a planar, 1-DoF (degree-of-freedom), closed chain has been chosen as said translational section of the Actuation System in order to obtain high rigidity, while the external body of the camera frame is used as a part of said translational section of the Actuation System.

FIG. 2 is an isometric view of a realistic embodiment of Armilleye adapted to orient a spindle. The end-effector of the 1-DoF (degree-of-freedom) mechanism, used as said translational section of the Actuation System, is a hollow socket hosting the spindle ; a motor fixed to the base provides torque to the spindle through a cardan joint and an extensible limb. In order to dispose the cardan joint with centre of rotation coincident to said spherical centre of rotation of the Actuation System, one extremity of said common fifth link of said spherical chains of the Actuation System has a fork shape.

FIG. 3 is an isometric representation of a realistic embodiment of said spherical section of the Actuation System adapted to orient two cameras like human eyes for stereoscopic vision with two view points (Armillhead).

FIG. 4 is an isometric representation of a realistic embodiment of said spherical section of the Actuation System adapted to carry and orient two end-effector wrists for manipulation, cooperative handling, machining or such similar applications.

FIG. 5 schematically represents the scanning methodology.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS FIG. 1 shows one preferred embodiment of the Actuation System object of the present invention. Due to the architecture of the mechanism that has been chosen for said translational section of the Actuation System, this particular embodiment of the Actuation System object of the present invention is also a particular embodiment of said Armilleye, further object of the present invention.

The first, second and third rotary actuators of the Armilleye, respectively 36,37, 38, are fixed to the base 0. The first spherical chain of the Armilleye is composed of the links: 6 (first), 12 (second link), 16 (third link), 8 (fourth link), 4 (fifth link). The second spherical chain of the Armilleye is composed of the links: 7 (first), 13 (second link), 17 (third link), 9 (fourth link), 4 (fifth link). Said link 4 is common to both two said spherical chains of the Armilleye. Said link 6 is connected to the shaft of said second rotary actuator 37 and can rotate with respect to said base 0 about the axis 2. Said link 7 is connected to the shaft of said first rotary actuator 36 and can rotate with respect to said base 0 about the axis 1. Said link 4 is connected to the shaft of said third rotary actuator 38 and can rotate with respect to the base about the axis 3. Said links 8 and 9 are connected to said link 4 by coaxial revolute joints, whose axes coincide with the axis 5 (Armilleye also includes embodiments such that these two revolute joints, respectively between said links 8 to 4 and 9 to 4, are not coaxial). Said link 6 is connected to said link 12 by a revolute joint, whose axis is 10, while said link 7 is connected to said link 13 by a revolute joint, whose axis is 11. Said link 12 is connected to said link 16 and said link 16 is connected to said link 8 by two coaxial revolute joints, whose axes coincide with the axis 14, which is the output axis of the first spherical chain of the Armilleye (Armilleye also includes embodiments such that said links 12 and 16 or 8 and 16 are rigidly connected together). Said link 13 is connected to said link 17 and said link 17 is connected to said link 9 by two coaxial revolute joints, whose axes coincide with the axis 15, which is the output axis of the second spherical chain of the Armilleye (Armilleye also includes embodiments such that said links 13 and 17 or 9 and 17 are rigidly connected together). Said axes 1, 2,3, 5,10, 11,14, 15, intersect at the spherical centre of rotation of the Armilleye 33.

The two said spherical chains of the Armilleye constitute the spherical section of the Armilleye. Said spherical section of the Armilleye is a 3-DoF (degree-of-freedom) mechanism, whose inputs are the angular orientation of said rotary actuators 36,37, 38. Said rotary actuators 36 and 37 command the angular orientation of said axes 14 and 15 about said axis 5, while said rotary actuator 38 commands the angular orientation of said axes 14 and 15 about said axis 3. Said axes 3,14, 15 define the horizon plane of the Armilleye, whose inclination with respect to said base 0 depends only on the angular orientation of said third rotary actuator 38. Outputs of said spherical section of the Armilleye are the orientation of the bisecting line of the planar angle formed by said axes 14, 15, about said spherical centre of rotation 33, and the span of such planar angle formed by said axes 14,15.

In the present embodiment, said translational section of the Actuation System consists of two closed, interconnected chains. The first chain consists of the links 32,22, 16,24, 35, while the second chain consists of the links 32,23, 17,25, 35.

Said link 22 is connected at one side to said link 16 by a revolute joint of axis 18; at the other side it is connected to said link 32 by a revolute joint of axis 28. Said link 24 is connected to said link 16 by a revolute joint of axis 20; it is connected at the other side to said link 35 by a revolute joint of axis 26. Said link 23 is connected at one side to said link 17 by a revolute joint of axis 19; at the other side it is connected to said link 32 by a revolute joint of axis 29. Said link 25 is connected to said link 17 by a revolute joint of axis 21; it is connected at the other side to said link 35 by a revolute joint of axis 27. Said axes 18,19, 20,21, 26,27, 28,29, are parallel themselves and are orthogonal to said horizon plane of the Armilleye (which is defined by said axes 3,14, 15). Said links 32 and 35 are connected by a cylindrical revolute joint of axis 34, in order that said link 35 can translate along said axis 34 with respect to said body 32. Said links 32 and 35 can also be connected by a prismatic joint of direction 34 without changing the behaviour of the mechanism. Said axis 34 passes through said spherical centre of rotation 33. Said axis 34 is the orientation axis of the Armilleye; due to the geometry of the present embodiment, said axis 34 is also the bisecting line of the planar angle formed by said axes 14,15.

The translational section of the Armilleye is a 1-DoF (degree-of-freedom) mechanism. Input coordinate of the translational section of the Armilleye is the span of the planar angle formed by said axes 14,15, while output freedom is a translation along said axis 34. The whole translational section of the Armilleye is suspended on said axes 14,15, and so oriented as a whole by the spherical section of the Armilleye.

The geometry of the links of the Armilleye in the present embodiment is such that the workspace is somewhat less than a full sphere; however, many applications require a smaller workspace.

FIG. 2 shows one preferred embodiment of Armilleye, according to FIG. 1, adapted to carry and orient a spindle.

The spindle 65 is connected to said link 32 by a revolute joint, whose axis coincides with said orientation axis of the Actuation System 34 and with the axis of rotation of said spindle 65. The shaft 63 is connected to said spindle 65 by a prismatic joint, whose sliding axis coincides with said orientation axis of the Actuation System 34, in order to transmit torque like an extensible limb. Said shaft 63 is connected to a cardan joint 61, whose centre of rotation coincides with said spherical centre of rotation 33. Said cardan joint 61 is connected at the opposite side to the rotary actuator 60, fixed to the base. The geometry of the other links of the Armilleye in the present embodiment is the same as in FIG. 1.

FIG. 3 shows one preferred embodiment of said spherical section of the Actuation System utilised as a device to support and orient two cameras for stereoscopic vision with two view points (Armillhead).

The embodiment of said spherical section of the Actuation System is the same as described with reference to FIG. 1. The first, second and third rotary actuators of the Armillhead, respectively 36,37, 38, are fixed to the base 0.

The first spherical chain of the Armillhead is composed of the links: 6 (first), 12 (second link), 16 (third link), 8 (fourth link), 4 (fifth link). The second spherical chain of the Armillhead is composed of the links: 7 (first), 13 (second link), 17 (third link), 9 (fourth link), 4 (fifth link). Said link 4 is common to both two said spherical chains of the Armillhead.

Said link 6 is connected to the shaft of said second rotary actuator 37 and can rotate with respect to said base 0 about the axis 2. Said link 7 is connected to the shaft of said first rotary actuator 36 and can rotate with respect to said base 0 about the axis 1. Said link 4 is connected to the shaft of said third rotary actuator 38 and can rotate with respect to the base about the axis 3. Said links 8 and 9 are connected to said link 4 by coaxial revolute joints, whose axes coincide with the axis 5 (Armillhead also includes embodiments such that these two revolute joints, respectively between said links 8 to 4 and 9 to 4, are not coaxial). Said link 6 is connected to said link 12 by a revolute joint, whose axis is 10, while said link 7 is connected to said link 13 by a revolute joint, whose axis is 11. Said link 12 is connected to said link 16 and said link 16 is connected to said link 8 by two coaxial revolute joints, whose axes coincide with the axis 14 (Armillhead also includes embodiments such that said links 12 and 16 or 8 and 16 are rigidly connected together). Said link 13 is connected to said link 17 and said link 17 is connected to said link 9 by two coaxial revolute joints, whose axes coincide with the axis 15 (Armillhead also includes embodiments such that said links 13 and 17 or 9 and 17 are rigidly connected together). Said axes 1,2, 3,5, 10,11, 14,15, intersect at the spherical centre of rotation of the Armillhead 33.

Each said spherical chain of the Armillhead carries a camera or other vision system, which is rigidly connected to its said fourth link. So, the first camera 39 of axis 41 is fixed to said link 8, while said camera 40 of axis 42 is fixed to said link 9. Said camera 39 is oriented with respect to said link 8 in order that said axis 41 forms a constant angle different from zero with the plane defined by said axes 14 and 5.

Analogously, said camera 40 is oriented with respect to said link 9 in order that said axis 42 forms with the plane defined by said axes 15 and 5 an angle equal and opposite to said angle formed by said axis 41 with the plane defined by said axes 14 and 5. Hence a value exists for the angle that said axis 14 forms with said axis 15, such that said axes 41 and 42 are parallel; in case that such angle between said axes 14 and 15 is smaller, then said axes 41 and 42 intersect in a point in front of the Armillhead. Due to kinematics reasons, said axes 41 and 42 are always in a plane parallel to the plane defined by said axes 3,14, 15. So, by suitably operating said rotary actuators 36,37, 38, it is possible to displace said point of intersection of said axes 41 and 42 on every object in the field of vision.

Then, by triangulation, it is possible to compute the 3D (three dimensional) location of every object in the field of vision.

FIG. 4 shows one preferred embodiment of said spherical section of the Actuation System utilised as a device to support and orient two end-effector wrists for manipulation, cooperative handling, machining or such similar applications.

The embodiment of said spherical section of the Actuation System is the same as described with reference to FIG. 1, FIG. 2 and FIG. 3. In place of two cameras or other vision systems, the spherical chains carry and orient two serial arms endowed with two grippers. As well as such arms, the spherical chains can carry any two or more, more or less complex systems.