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.

An apparatus and a method for planning contact-interaction trajectories are provided. The apparatus is a robot that accepts contact interactions between the robot and the environment. The robot stores a dynamic model representing geometric, dynamic, and frictional properties of the robot and the environment, and a relaxed contact model to representing dynamic interactions between the robot and the object via virtual forces. The robot further determines, iteratively until a termination condition is met, a trajectory, associated control commands for controlling the robot, and virtual stiffness values by performing optimization reducing stiffness of the virtual force and minimizing a difference between the target pose of the object and a final pose of the object moved from the initial pose. Further, an actuator moves a robot arm of the robot according to the trajectory and the associated control commands.

A method for planning a manipulation task of an agent, particularly a robot. The method includes: learning a number of manipulation skills wherein a symbolic abstraction of the respective manipulation skill is generated; determining a concatenated sequence of manipulation skills selected from the number of learned manipulation skills based on their symbolic abstraction so that a given goal specification indicating a given complex manipulation task is satisfied; and executing the sequence of manipulation skills.

Example implementations may relate to methods and systems for determining a safe trajectory for movement of an object by a robotic system. According to these various implementations, the robotic system may determine at least first and second candidate trajectories for moving the object. For at least a first point along the first candidate trajectory, the robotic system may determine a predicted cost of dropping the object at the first point along the first candidate trajectory. And for at least a second point along the second candidate trajectory, the robotic system may determine a predicted cost of dropping the object at the second point along the second candidate trajectory. Then, based on these various determined predicted costs, the robotic system may select between the first and second candidates trajectories and may then move the object along the selected trajectory.

Dynamic control of a center of mass position is based on replacement of discrete motion of macro body (counterweighing solid or counterbalancing mechanisms) for continuous molecular flow of counterweighing liquid. Redistributing liquid counterweight between chambers attached to independently moving parts of robot allows its motion to new stable position without disruption in static stability and dynamic balance. Various embodiments include bipods/humanoids, wheeled locomotion robots and hybrid wheeled/multi-pod bio-like robotic systems; some embodiments allow reversible mutual reconfiguration between various structural arrangements. In humanoid embodiments, method allows moving on uneven terrain or ascending staircases while maintaining static stability; method also decreases the probability of fall and secures self-rising if a fall occurred. In some embodiments liquid counterweight may be transferred upon high barriers exceeding the height of robot by a few folds, such as walls of the building or ledge or steep slope in mountains, thus providing robots with capability principally not available to prior art.

Methods, apparatus, systems, and computer-readable media are provided for determining one or more spatial constraints associated with an object to be acted upon by a robot; determining a plurality of candidate physical arrangements of the object that satisfy the one or more spatial constraints; calculating, for one or more of the plurality of candidate physical arrangements of the object, a candidate physical arrangement cost that would be incurred as a result of the robot acting upon the object in the candidate physical arrangement; and selecting, from the plurality of candidate physical arrangements, a candidate physical arrangement associated with a candidate physical arrangement cost that satisfies a criterion.

An apparatus and a method for planning contact-interaction trajectories are provided. The apparatus is a robot that accepts contact interactions between the robot and the environment. The robot stores a dynamic model representing geometric, dynamic, and frictional properties of the robot and the environment, and a relaxed contact model to representing dynamic interactions between the robot and the object via virtual forces. The robot further determines, iteratively until a termination condition is met, a trajectory, associated control commands for controlling the robot, and virtual stiffness values by performing optimization reducing stiffness of the virtual force and minimizing a difference between the target pose of the object and a final pose of the object moved from the initial pose. Further, an actuator moves a robot arm of the robot according to the trajectory and the associated control commands.

A method for determining a trajectory of a robot from a starting position to a target position is provided. The starting position and the target position are manually defined by a user in a real environment of the robot. Then a collision-free trajectory of the robot from the starting position to the target position is determined, based on the surroundings of the robot. Also provided is a device, a robot system, a computer program and a machine-readable storage medium.

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.

To generate a more appropriate path, provided is a robot path generation device including circuitry configured to: hold a track planning module learning data set, in which a plurality of pieces of path data generated based on a motion constraint condition of a robot, and evaluation value data, which corresponds to each of the plurality of pieces path data and is a measure under a predetermined evaluation criterion, are associated with each other; and generate, based on a result of a machine learning process that is based on the track planning module learning data set, a path of the robot between a set start point and a set end point, which are freely set.