I want to simultaneously improve robot's manipulation dexterity and reliability. I am currently exploring the following aspects of manipulation:
1. Simple and reliable end-effector hardware;
2. A deep understanding of the mechanics of manipulation;
3. Clever control strategy to fight the inevitable uncertainties in the reality.
In hybrid force-velocity control, the robot can use velocity control in some directions to follow a trajectory, while performing force control in other directions to maintain contacts with the environment regardless of positional errors. We call this way of executing a trajectory hybrid servoing. We propose an algorithm to compute hybrid force-velocity control actions for hybrid servoing. We quantify the robustness of a control action and make trade-offs between different requirements by formulating the control synthesis as optimization problems. Our method can efficiently compute the dimensions,directions and magnitudes of force and velocity controls. We demonstrated by experiments the effectiveness of our method in several contact-rich manipulation tasks.
We consider reorienting 3D objects on a table using a two-finger pinch gripper. Given the 3D mesh model of the object, our algorithm solves for the gripper motions that are required to transit between arbitrary object poses, grasping positions and gripper poses. The two motion primitives we used, pivoting and compliant rolling, enable us to decompose the planning problem and solve it more efficiently. Our algorithm can work with approximated (simplified) mesh models while being robust to approximation errors, thereby allowing us to efficiently handle object shapes with originally thousands of facets. We show the effectiveness of the proposed method by testing on objects with non-trivial geometry in both simulations and experiments. Results show that our algorithm can solve a larger range of reorienting problems with less number of making and breaking contacts when compared to traditional pick-and-place based methods, especially when the gripper workspace is highly constrained.
In this project, we investigate the planar dynamic pivoting problem, in which a pinched object is reoriented to a desired pose through wrist swing motion and grip force regulation. Traditional approaches based on friction compensation do not work well for this problem, as we observe the torsional friction at the contact has large uncertainties during pivoting. In addition, the discontinuities of friction and the lower bound constraint on the grip force all make dynamic pivoting a challenging task for robots. To address these problems, we propose a robust control strategy that directly uses friction as a key input for dynamic pivoting, and show that active friction control by regulating the grip force significantly improves system stability. In particular, we embed a Lyapunov-based control law into a quadratic programming framework, which also ensures real-time computational speed and the existence of a solution. The proposed algorithm has been validated on our dynamic pivoting robot that emulates human wrist-finger configuration and motion. The object orientation can quickly converge to the target even under considerable uncertainties from friction and object grasping position, where traditional methods fail.
We propose a framework for performing single contact point planar pushing with unknown pressure distribution. The problem is challenging due to the stochastic and under-actuated system properties. Using as few as two data points, our method can estimate the control-related model parameters. To achieve posture stabilization, we use differential dynamic programming and re-plan with updated model when large deviation occurs. We demonstrate empirical success in posture stabilization and robustness with respect to different pressure distributions.
We present an algorithm for obtaining an optimal control policy for hybrid dynamical systems in cluttered environments. To the best of our knowledge, this is the first attempt to have a locally optimal solution for this specific problem setting. Our approach extends an optimal control algorithm for hybrid dynamical systems in the obstacle-free case to environments with obstacles. Our method does not require any preset mode sequence or heuristics to prune the exponential search of mode sequences. By first solving the relaxed problem of getting an obstacle-free, dynamically feasible trajectory and then solving for both obstacle-avoidance and optimality, we can generate smooth, locally optimal control policies. We demonstrate the performance of our algorithm on a box-pushing example in a number of environments against the baseline of randomly sampling modes and actions with a Kinodynamic RRT.
I worked on bipedal walking robot during undergrad in Tsinghua. I lead a 15-students team, spent a year in developing an autonomous soccer-playing humanoid robot control system based on ROS. I was responsible for motion control module and overall team management. We got 3rd place in Robocup 2013(Netherlands) and Robocup 2014 (Brazil).