The Mass Matrix and the Mug, Why a Robot Slows Down When It Picks Something Up
9 hours ago
- #adaptive control
- #robotics dynamics
- #mass matrix
- Explains the mass matrix M(θ), a 7x7 grid describing how heavy a robot arm feels in a specific pose with a load, which changes when picking up an object.
- Describes dynamics in robotics: calculating required motor torques for joint angles, contrasting with kinematics for motion planning.
- Highlights inertia term in the manipulator equation where the mass matrix resides, showing diagonal entries as joint inertia and off-diagonal as coupling.
- Notes how adding a 1.5kg mug significantly alters the mass matrix, doubling wrist inertia and affecting joint coordination.
- Mentions adaptive control methods, like Slotine and Li's 1987 approach, and a 2026 paper estimating payload for real-time updates.
- Discusses recent advancements: using natural dynamics for faster bipedal locomotion, actuator identification without sensors, and open-source tools like Drake.
- Emphasizes the importance of torque control over position control for cobots and humanoids, requiring real-time mass matrix updates for tasks like handling humans or fragile objects.