At the hardware level, servo drives are traditionally controlled in the position or velocity domain. Frequently, roboticists do not recognize the significance of selecting one control domain over the other. This paper argues that velocity control is the clear choice for reactive trajectory modification and online motion planning. Velocity control provides smoother motion, is more suitable for low control frequencies, more robust to control signal variation, and it reduces collision forces more effectively. The primary advantage of position control is inherent safety with regards to signal delays. To support these points, this paper presents mathematical, graphical and experimental evidence. New experimental results explore the robustness of velocity control with regards to low-frequency control signals and signal delay. The experiment is based on a cooperative manipulation task performed with stiff industrial manipulators. Position and velocity control are also directly compared using a one-DOF linear actuator force control task using total work performed as a performance metric.

The development of control strategies that allow stiff industrial robots to operate safely in unstructured environments is a significant challenge. This paper integrates two strategies that improve safety for industrial manipulators in uncertain conditions. First, software compliance in the task space is implemented using force feedback. End-effector compliance is vital for many tasks, such as interacting with humans or manipulating uncertain payloads. Beyond compliance, a collision detection algorithm detects collisions based on joint torque deviation from a dynamic model. Collisions can be detected at any point along the manipulator via loading or impulse anomalies. Joint torque data is typically noisy, and the accuracy of the robot dynamic model is limited, so an Extended Kalman Filter (EKF) was developed to improve the torque estimates. Experiments and demonstrations were performed using a commercially available 7DOF industrial robot. The EKF improved collision detection during unplanned contact tasks, and the method described here is hardware agnostic and extensible.

If a Lyapunov function is known, a dynamic system can be stabilized. However, computing a Lyapunov function is often challenging. This paper takes a new approach; it assumes a basic Lyapunov-like function then seeks to numerically diminish the Lyapunov value. If the control effort would have no effect at any iteration, the Lyapunov-like function is switched in an attempt to regain control. The method is tested on four simulated systems to give some perspective on its usefulness and limitations. A highly coupled 3rd order system demonstrates the approach’s general applicability and finally the coordinated control of 7 motors for a robotic application is considered. Details on the publicly available software packages for application agnostic software and hardware environments are also presented.

No abstract

If a Lyapunov function is known, a dynamic system can be stabilized. However, the search for a Lyapunov function is often challenging. This paper takes a new approach to avoid such a search; it assumes a basic Control Lyapunov Function [CLF] then seeks to numerically diminish the value of the Lyapunov function. If a singularity arises during calculations with the default CLF, a complementary function is used. The complementary function eliminates a common cause of singularities with the default CLF. While many other algorithms from the literature use switched or time-varying CLF’s, the presented method is unique in that the CLF’s do not require prior calculation and the technique applies globally. The method is proven and demonstrated for SISO systems in normal form and then demonstrated on a higher-order system of a more general type.