Proprioceptive Sensing for a Legged Robot

This thesis provides a methodology of sensory system development for a hexapod robot, working toward the development of dynamic behaviors utilizing feedback controllers. We develop an approach to utilizing strain gauges together with a simple data driven phenomenological model that simultaneously delivers information of leg touchdown, leg configuration, and ground reaction force suitable for robots with compliant legs. The strain gauges are implemented and models are constructed on two versions of RHex robot compliant legs. Leg configuration is further evaluated under realistic robot operating conditions by means of a high speed visual ground truth system. We then introduce a continuous time 6 degree of freedom (DOF) body pose estimator for a walking hexapod robot. Our algorithm uses six leg configurations together with prior knowledge of the ground and robot kinematics to compute instantaneous estimates of the body pose. We implement this estimation procedure on RHex and evaluate the performance of this algorithm at widely varying body speeds and over dramatically different ground conditions by means of a 6 DOF vision-based ground truth measurement system (GTMS). We also compare the odometry performance to that of sensorless schemes—both legged as well as on a wheeled version of the robot—using GTMS measurements of traversed distance. Finally, we report on a hybrid 12-dimensional full body state estimator for a jogging hexapod robot on level terrain with regularly alternating ground contact and aerial phases of motion. We use a repeating sequence of dynamical models switched in and out of an Extended Kalman Filter to fuse measurements from a body pose sensor and inertial sensors. Our inertial measurement unit supplements the traditionally paired 3-axis gyroscope/accelerometer with a set of three additional 3-axis accelerometer suites, thereby providing additional angular acceleration measurement (inertia torque), avoiding the need for localization of the accelerometer at the center of mass on the robot’s body, and simplifying installation and calibration. We implement this estimation procedure offline, using data extracted from numerous repeated runs of RHex and evaluate its performance with reference to GTMS, also comparing the relative performance of different fusion approaches implemented via different model sequences.