Hybrid mechanical systems arise in many applications, including hopping, walking, and climbing robots where contact with the ground changes; skid-steering vehicles where Coulomb friction introduces stick/slip behavior; and prosthetic devices that interact with objects. All of these applications are influenced by nonlinearity, nonsmooth transitions, and uncertainty, and these systems demand new tools in motion planning and control.
This work takes advantage of the specific structure of mechanical systems to bound the propagation of uncertainty and to develop feedback controllers that maximize robustness of execution. The work builds on state-of-the-art techniques in motion planning and estimation, including sample-based and optimization-based planning, leading to tools for uncertain hybrid mechanical systems that are analogous to control and estimation tools used for linear systems.
Example systems are used to drive algorithm development as well as to verify performance. These include 1) the Monkeybot, a robot that uses electromagnets and a single motor to locomote along a vertical wall; 2) a parkour robot that uses mechanical contact and jumping to climb narrow passages; and 3) a skid-steered vehicle that experiences discontinuous dynamics due to stick/slip friction effects. All phases of this work include participation of undergraduates and minority students. In addition to dissemination in conferences and journals, results are disseminated on a publicly viewable wiki.
Creating robots that robustly walk in environments with varying terrain, poor visibility, and changing environmental conditions continues to be a grand challenge in robotics. All these types of uncertainties need to be incorporated into a robot’s decision making, and this project has analyzed planning and control for robots (and examples such as vehicles) that experience impacts and other forms of discrete changes in uncertain environments. This work has had high public visibility. As part of this project, the PI and the students involved created an introductory systems analysis course on the Coursera platform. Over 30,000 students signed up for the two offerings of the free class in autumn, 2013 and spring, 2014. This class is at www.coursera.org/course/modelsystems. Moreover, the PIs’ laboratory created a display at the Museum of Science and Industry, Chicago during National Robotics Week (2012-2014). Thousands of elementary and secondary school children interacted with the PIs’ robots and learned about the role of control and software in embedded systems. One of the major developments in this project was the development of a new robot. The Gibbot is a bio-inspired climbing robot that swings from one handhold to the next. The robot serves as a model system in our work on planning and state estimation of hybrid mechanical systems. In addition to serving as a research platform, the Gibbot will be included in the Museum of Science and Industry, Chicago. This world class science museum is launching a permanent robotics exhibit and our robot has been invited to be a part of the permanent display. Students have worked in several teams over many months on various technical challenges associated with building a robot intended to consistently perform 7 days a week. Some students worked on mechanical features of the robot, including building the links of the robot (which house all of the actuators, electronics, and batteries), sizing the motor for the robot, charging the robot after a performance with no human intervention through the links, and reducing the amount of friction between the electromagnetic hands and the steel wall that serves as the Gibbot’s environment and stage. Other students worked on the electronics of the robot for the microcontroller, battery charger, and onboard sensors, as well as building a prototype for an electro-permanent magnet as a failsafe against power loss. Finally, another group of students worked on tracking the robot with a vision system and implementing higher-level motion planning and control strategies for the robot on a PC. The work can be found online in a world-readable Github repository, https://github.com/nxrlab/gibbot, and highlights can be found at http://nxr.northwestern.edu/research/dynamic-locomotion/gibbot. The project included analysis of vehicle driving styles, focusing on how a vehicle can be driven when traction is uncertain. The result of this part of the effort was that many things we learn when driving, such as releasing the brakes momentarily when hitting a patch of ice, are optimal responses to uncertainty about traction. A major result of this work includes illustrating how to design algorithms that automatically respond to uncertainty, providing a prototype for how advanced software assistance in vehicles can be designed. Papers from both projects can be found at http://nxr.northwestern.edu.
