Generating a robot control policy that regulates both motion control and interaction with an environment and/or includes a learned potential function and/or dissipative field. Some implementations relate to resampling temporally distributed data points to generate spatially distributed data points, and generating the control policy using the spatially distributed data points. Some implementations additionally or alternatively relate to automatically determining a potential gradient for data points, and generating the control policy using the automatically determined potential gradient. Some implementations additionally or alternatively relate to determining and assigning a prior weight to each of the data points of multiple groups, and generating the control policy using the weights. Some implementations additionally or alternatively relate to defining and using non-uniform smoothness parameters at each data point, defining and using d parameters for stiffness and/or damping at each data point, and/or obviating the need to utilize virtual data points in generating the control policy.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for avoiding singular configurations of a robot. A singular configuration of the robot is obtained. A location of an end effector of the robot when the robot is in the singular configuration is determined. For each of a plurality of voxels in a workcell, a distance from the voxel to the location of the end effector when the robot is in the singular configuration is computed. A negative potential gradient of the computed distance is computed. Control rules are generated, wherein the control rules, when followed by the robot, offset the trajectory of the robot according to the negative potential gradient.

Generating a robot control policy that regulates both motion control and interaction with an environment and/or includes a learned potential function and/or dissipative field. Some implementations relate to resampling temporally distributed data points to generate spatially distributed data points, and generating the control policy using the spatially distributed data points. Some implementations additionally or alternatively relate to automatically determining a potential gradient for data points, and generating the control policy using the automatically determined potential gradient. Some implementations additionally or alternatively relate to determining and assigning a prior weight to each of the data points of multiple groups, and generating the control policy using the weights. Some implementations additionally or alternatively relate to defining and using non-uniform smoothness parameters at each data point, defining and using d parameters for stiffness and/or damping at each data point, and/or obviating the need to utilize virtual data points in generating the control policy.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for avoiding singular configurations of a robot. A singular configuration of the robot is obtained. A location of an end effector of the robot when the robot is in the singular configuration is determined. For each of a plurality of voxels in a workcell, a distance from the voxel to the location of the end effector when the robot is in the singular configuration is computed. A negative potential gradient of the computed distance is computed. Control rules are generated, wherein the control rules, when followed by the robot, offset the trajectory of the robot according to the negative potential gradient.

The invention relates to a robot, a robot control system, and a method for controlling a robot. The robot comprises a movable, multi-membered robot structure (102) that can be driven by means of actuators (101), at least one marked structural element S being defined on the movable robot structure (102), with at least one point P.sub.S marked on the structural element S. The robot is designed such that, in an input mode, it learns positions POS.sub.PS of the point PS and/or poses of the structural element S in a work space of the robot, the user exerting an input force F.sub.EING on the movable robot structure in order to move the structural element S, which is conveyed to the point P.sub.S as F.sub.EING,PS, and/or to the structural element S as torque M.sub.EING,S. A control device (103) of the robot is designed such that, in the input mode, the actuators (101) are controlled on the basis of a pre-defined space-fixed virtual 3D grid that at least partially fills the work space, such that the structural element S is moved with a pre-defined force F.sub.GRID (POS.sub.PS), according to the current position POS.sub.PS of the point P.sub.S in the 3D grid, to the adjacent grid point of the 3D grid or in a grid point space defined around the adjacent grid point of the 3D grid, the point P.sub.S of the structural element S remaining on said adjacent grid point or in said grid point space in the event of the following holding true: |F.sub.EING,PS|<|F.sub.GRID(POS.sub.PS) and/or, in the input mode, the actuators (101) are controlled on the basis of a pre-defined virtual discrete 3D orientation space O, where the 3D orientation space O=: (α.sub.i, β.sub.j, γ.sub.k) where i=1, 2, . . . , I, j=1, 2, . . . J, k=1, 2, . . . , K is defined or can be defined by a pre-defined angle α.sub.i, β.sub.j, γ.sub.k, in such a way that the structural element S is moved with a pre-defined torque)(SO ROM according to the current orientation OR.sub.S of the structural element, towards the adjacent discrete orientation of the 3D orientation space O=: (α.sub.i, β.sub.j, γ.sub.k), S, the structural element remaining in said adjacent discrete orientation of the 3D orientation space O in the event that the following holds true: |M.sub.EING,S|<|M.sub.O(OR.sub.S).

Generating a robot control policy that regulates both motion control and interaction with an environment and/or includes a learned potential function and/or dissipative field. Some implementations relate to resampling temporally distributed data points to generate spatially distributed data points, and generating the control policy using the spatially distributed data points. Some implementations additionally or alternatively relate to automatically determining a potential gradient for data points, and generating the control policy using the automatically determined potential gradient. Some implementations additionally or alternatively relate to determining and assigning a prior weight to each of the data points of multiple groups, and generating the control policy using the weights. Some implementations additionally or alternatively relate to defining and using non-uniform smoothness parameters at each data point, defining and using d parameters for stiffness and/or damping at each data point, and/or obviating the need to utilize virtual data points in generating the control policy.

Generating a robot control policy that regulates both motion control and interaction with an environment and/or includes a learned potential function and/or dissipative field. Some implementations relate to resampling temporally distributed data points to generate spatially distributed data points, and generating the control policy using the spatially distributed data points. Some implementations additionally or alternatively relate to automatically determining a potential gradient for data points, and generating the control policy using the automatically determined potential gradient. Some implementations additionally or alternatively relate to determining and assigning a prior weight to each of the data points of multiple groups, and generating the control policy using the weights. Some implementations additionally or alternatively relate to defining and using non-uniform smoothness parameters at each data point, defining and using d parameters for stiffness and/or damping at each data point, and/or obviating the need to utilize virtual data points in generating the control policy.

Generating a robot control policy that regulates both motion control and interaction with an environment and/or includes a learned potential function and/or dissipative field. Some implementations relate to resampling temporally distributed data points to generate spatially distributed data points, and generating the control policy using the spatially distributed data points. Some implementations additionally or alternatively relate to automatically determining a potential gradient for data points, and generating the control policy using the automatically determined potential gradient. Some implementations additionally or alternatively relate to determining and assigning a prior weight to each of the data points of multiple groups, and generating the control policy using the weights. Some implementations additionally or alternatively relate to defining and using non-uniform smoothness parameters at each data point, defining and using d parameters for stiffness and/or damping at each data point, and/or obviating the need to utilize virtual data points in generating the control policy.

Generating a robot control policy that regulates both motion control and interaction with an environment and/or includes a learned potential function and/or dissipative field. Some implementations relate to resampling temporally distributed data points to generate spatially distributed data points, and generating the control policy using the spatially distributed data points. Some implementations additionally or alternatively relate to automatically determining a potential gradient for data points, and generating the control policy using the automatically determined potential gradient. Some implementations additionally or alternatively relate to determining and assigning a prior weight to each of the data points of multiple groups, and generating the control policy using the weights. Some implementations additionally or alternatively relate to defining and using non-uniform smoothness parameters at each data point, defining and using d parameters for stiffness and/or damping at each data point, and/or obviating the need to utilize virtual data points in generating the control policy.

The invention relates to a robot, a robot control system, and a method for controlling a robot. The robot comprises a movable, multi-membered robot structure (102) that can be driven by means of actuators (101), at least one marked structural element S being defined on the movable robot structure (102), with at least one point P.sub.S marked on the structural element S. The robot is designed such that, in an input mode, it learns positions POS.sub.PS of the point PS and/or poses of the structural element S in a work space of the robot, the user exerting an input force F.sub.EING on the movable robot structure in order to move the structural element S, which is conveyed to the point P.sub.S as F.sub.EING,PS, and/or to the structural element S as torque M.sub.EING,S. A control device (103) of the robot is designed such that, in the input mode, the actuators (101) are controlled on the basis of a pre-defined space-fixed virtual 3D grid that at least partially fills the work space, such that the structural element S is moved with a pre-defined force F.sub.GRID (POS.sub.PS), according to the current position POS.sub.PS of the point P.sub.S in the 3D grid, to the adjacent grid point of the 3D grid or in a grid point space defined around the adjacent grid point of the 3D grid, the point P.sub.S of the structural element S remaining on said adjacent grid point or in said grid point space in the event of the following holding true: |F.sub.EING,PS|<|F.sub.GRID(POS.sub.PS) and/or, in the input mode, the actuators (101) are controlled on the basis of a pre-defined virtual discrete 3D orientation space O, where the 3D orientation space O=: (.sub.i, .sub.j, .sub.k) where i=1, 2, . . . , I, j=1, 2, . . . J, k=1, 2, . . . , K is defined or can be defined by a pre-defined angle .sub.i, .sub.j, .sub.k, in such a way that the structural element S is moved with a pre-defined torque)(SO ROM according to the current orientation OR.sub.S of the structural element, towards the adjacent discrete orientation of the 3D orientation space O=: (.sub.i, .sub.j, .sub.k), S, the structural element remaining in said adjacent discrete orientation of the 3D orientation space O in the event that the following holds true: |M.sub.EING,S|<|M.sub.O(OR.sub.S).