humanoidsdata.com

Search

Search companies, datasets, articles, and glossary terms for humanoids and embodied AI.

← All glossary terms

Hardware & control

Robot dynamics

Robot dynamics describes the relationship between forces and torques acting on a robot and the motion they produce, accounting for mass, inertia, gravity, velocity-dependent effects and external contact. Forward dynamics predicts acceleration from applied forces; inverse dynamics computes forces or torques for a specified motion.

Also known as: robotic dynamics

Updated

Dynamics adds the causes of motion

Robot kinematics relates joint values to pose and velocity without asking what caused them. Dynamics adds link mass and inertia, applied force and torque, gravity, friction, compliance and environmental contact. The result is an equation of motion that constrains which trajectories a physical robot can produce.

Modern Robotics Chapter 8 presents the standard rigid-body terms: a configuration-dependent mass matrix, velocity-product effects such as Coriolis and centripetal terms, gravity and external wrenches. Real mechanisms can also require models of motor inertia, gearing, damping, backlash and flexible structures.

Forward and inverse dynamics answer opposite questions

Forward dynamics starts with robot state and applied forces or torques, then calculates acceleration. Integrating those accelerations produces a simulated trajectory. Inverse dynamics starts with state and a requested acceleration, then calculates the joint forces or torques needed to realise it.

The Springer Handbook treatment covers both problems for general robot mechanisms. Inverse dynamics is useful for model-based control, but its answer is only physically achievable when actuator, contact and friction constraints are satisfied.

Humanoids add floating-base and contact dynamics

A humanoid’s root body is not fixed to the world and its contacts change as it walks or manipulates. Ground and object contact forces therefore participate directly in the dynamics. Impacts can cause velocity changes that a smooth kinematic trajectory does not represent.

Dynamic data should identify the robot model and inertial parameters, coordinate and sign conventions, applied and measured torque, external wrench, contact state, payload, timestep and solver assumptions. Simulator state is not automatically hardware ground truth: inaccurate friction, compliance or actuator models can produce motion that the real robot cannot match.

Sources