For safety-critical robotic systems, Control Barrier Functions (CBFs) provide a solid mathematical way to ensure that a system state always remains within a desired safe set. Usually, they are implemented as a secondary controller that alters the control input of the base controller when necessary. However, when naive CBFs are used to enforce positional safety constraints, they can only regulate the joint velocities and therefore have to rely on lower-level velocity controllers. The tracking capabilities of these nested controllers are limited by the system inertia, and may therefore allow the system to enter the unsafe zone. [1, 2]
Solutions that circumvent this problem are presented in the form of energy-based CBFs, a term originally coined by Singletary et al.[2]. This flavour of CBF uses an energy-based safe-set that takes into account the underlying system dynamics by incorporating the nested velocity controller dynamics. Different forms have been proposed over the years, for instance, passivity-preserving methods that utilize e.g. energy tanks [3] and damping injection [4]. A recent contribution by Califano et al. exploits the combination of CBFs and a base controller that already preserves passivity. Here, conditions are found under which passivity is preserved without degrading the base controller performance. [5]
However, experimental validation has not yet been achieved without relying on nested controllers to handle lower-level dynamics. Also, the focus has mostly been on collision avoidance instead of physical interaction. In this thesis an energy-based CBF control system will be implemented on a physical robotic manipulator that directly controls actuator torque, to fill the gap between theory and practice. Experiments will be conducted to validate safety performance for (unforeseen) interactions.
As far as time allows, multiple extensions are going to be made to this framework. These include methods to modulate robot momentum towards human operators and power transfer limitations.
[1] F. Ferraguti, C. T. Landi, A. Singletary, H.-C. Lin, A. Ames, C. Secchi, and M. Bonf`e, “Safety and Efficiency in Robotics: The Control Barrier Functions Approach,” pp. 15–30, 2015.
[2] A. Singletary, S. Kolathaya, and A. D. Ames, “Safety-Critical Kinematic Control of Robotic Systems,” Proceedings of the American Control Conference, vol. 2021-May, pp. 14–19, 2021.
[3] B. Capelli, C. Secchi, and L. Sabattini, “Passivity and Control Barrier Functions: Optimizing the Use of Energy,” IEEE Robotics and Automation Letters, vol. 7, pp. 1356–1363, 4 2022.
[4] G. Notomista, X. Cai, J. Yamauchi, and M. Egerstedt, “Passivity-Based Decentralized Control of Multi-Robot Systems With Delays Using Control Barrier Functions,” in 2019 International Symposium on Multi-Robot and Multi-Agent Systems (MRS), pp. 231–237, IEEE, 8 2019.
[5] F. Califano, “Passivity-Preserving Safety-Critical Control Using Control Barrier Functions,” IEEE Control Systems Letters, vol. 7, pp. 1742–1747, 2023.