Robotics
Contact-rich manipulation
Definition
Contact-rich manipulation is robotic manipulation in which making, maintaining, breaking, or sliding contacts is central to completing the task. The robot must reason about forces and changing contact modes, not only collision-free motion.
Also known as: Contact-rich robot manipulation
Updated
Contact is part of the task
Collision-free motion planning normally treats contact with obstacles as something to avoid. Contact-rich manipulation instead relies on contact. Peg insertion, pushing, sliding, cap screwing, and using an environmental support all require the robot to enter and leave contact modes while controlling motion and force.
Levine and colleagues demonstrated learned policies for tasks including tight-fitting assembly, ring placement, shoe-tree insertion, and bottle-cap screwing. These experiments show examples of contact-rich skills; they do not define a single algorithm that all such tasks must use.
Why contact changes planning and control
Small changes in pose can change which surfaces touch and whether they stick, slide, or separate. Rigid contact models can therefore produce discontinuous dynamics and many possible mode sequences. Pang and colleagues identify the growth in contact modes, non-smooth dynamics, and non-convex planning as central difficulties, then study smoothing within a quasi-dynamic model.
A system may combine force control, compliant motion, tactile sensing, trajectory optimisation, or learned policies. Contact-implicit optimisation is one planning approach; contact-rich manipulation is the broader task setting.
Successful contact is not automatically safe
Simulation can approximate friction, impact, and deformation differently from the physical system. Sensors have noise and limited bandwidth, and a small calibration error can turn a desired sliding contact into jamming. Learned policies also need enough variation in objects, poses, and contact conditions to support claims of generalisation.
Completion rate alone does not describe contact quality. Force peaks, damage, recovery after a missed contact, execution time, and behaviour outside the training distribution can matter as much as whether the final object pose was reached.
Sources
Related terms
Robotic manipulation
Robotic manipulation is the use of a robot to change an object's position, orientation, or state through physical interaction. It includes grasping and moving objects as well as actions such as pushing or carrying them without a grasp.
Force control
Force control regulates the force or wrench a robot applies to its environment. It may use a robot model, measured interaction forces, or both to produce joint commands that achieve a desired contact load.
Contact-implicit optimization
Contact-implicit optimization plans motion while allowing contact events and forces to emerge from contact constraints in the optimization. It avoids requiring every contact transition to be fixed in a predefined mode sequence.
Tactile sensing
Tactile sensing measures information arising from physical contact, such as contact geometry, deformation, or force. Robots use it to observe interactions at their fingers, grippers, feet, or other contact surfaces.