Sensing the forces applied to the palm of a robot manipulator requires robust compliant sensors guarded against collision. The force sensing palm for the MLab Simple Hand measures three components of net contact force in a simple robust design using flexure springs and optical position sensing. The hand is calibrated in an automatic process. Measurements are included to compare performance to the theoretical model.

Manipulators with compliant actuation exhibit passive joint displacements when exposed to external forces or collisions. This paper demonstrates that this displacement information is sufficient to infer a coarse estimate of the location of an incidental collision. Three techniques for contact point detection are compared: a closed-form inference model based on a serial chain with joint springs, a variation on Self Posture Changeability, and an empirical memory-based model of joint trajectories. The methods were experimentally tested using a Shadow Hand on an industrial Motoman SDA10 arm to quantify localization performance, actively discover and avoid a thin obstacle and localize and grasp a cup.

We propose a model-based reinforcement learning algorithm for biped walking in which the robot learns to appropriately modulate an observed walking pattern. Via-points are detected from the observed walking trajectories using the minimum jerk criterion. The learning algorithm modulates the via-points as control actions to improve walking trajectories. This decision is based on a learned model of the Poincaré map of the periodic walking pattern. The model maps from a state in the single support phase and the control actions to a state in the next single support phase. We applied this approach to both a simulated robot model and an actual biped robot. We show that successful walking policies are acquired.

We explore the use of computational optimal control techniques for automated construction of policies in complex dynamic environments. Our implementation of dynamic programming is performed in a reduced dimensional subspace of a simulated four-DOF biped robot with point feet. We show that a computed solution to this problem can be generated and yield empirically stable walking that can handle various types of disturbances.

We propose a model-based reinforcement learning algorithm for biped walking in which the robot learns to appropriately place the swing leg. This decision is based on a learned model of the Poincare map of the periodic walking pattern. The model maps from a state at the middle of a step and foot placement to a state at next middle of a step. We also modify the desired walking cycle frequency based on online measurements. We present simulation results, and are currently implementing this approach on an actual biped robot.

We developed a robust control policy design method in high-dimensional state space by using differential dynamic programming with a minimax criterion. As an example, we applied our method to a simulated five link biped robot. The results show lower joint torques from the optimal control policy compared to a hand-tuned PD servo controller. Results also show that the simulated biped robot can successfully walk with unknown disturbances that cause controllers generated by standard differential dynamic programming and the hand-tuned PD servo to fail. Learning to compensate for modeling error and previously unknown disturbances in conjunction with robust control design is also demonstrated. We also applied proposed method to a real biped robot for optimizing swing leg trajectories.

The bow leg hopper is a novel locomotion design with a highly resilient leg that resembles an archer's bow. During the flight a "thrust" actuator adds elastic energy to the leg, which is automatically released during stance to control hopping height. Lateral motion is controlled by directing the leg angle at touchdown, which determines the angle of takeoff or reflection. The leg pivots freely on a hip bearing, and is automatically decoupled from the leg-angle positioner during stance to preclude hip torques that would disturb body attitude. Upright attitude is maintained without active control by allowing the body to "hang" from the hip joint. Preliminary experiments with a planar prototype have demonstrated impressive performance, high efficiency, and low power requirements. Current experiments are focused on developing control and planning schemes to enable locomotion over discrete "stepping stones" and obstacles.

The bow leg hopper is a new design for a locomotion system with a resilient, flexible leg. It features a passive stance phase and natural pitch stability. It is controlled by actuators that configure the leg angle and stored leg energy during flight. During the stance, the actuators are mechanically decoupled from the leg and the stored energy is released. The trajectory is determined by the spring-mass physics and the state of the leg at impact. This design casts the controller as a function mapping three trajectory parameters to two control outputs once every hopping cycle. Our particular solution uses a combination of graph-search planning and feedback control. The planner searches the sequences of foot placements and computes control outputs using numerical solution of a physical model. The feedback control is computed once per bounce. Experimental data from a planar prototype are included demonstrating navigation of simple artificial terrain.