Principle:Google deepmind Mujoco Passive Forces
| Knowledge Sources | Domains | Last Updated |
|---|---|---|
| Google DeepMind MuJoCo | Physics Simulation | 2025-02-15 |
Overview
Description: Passive forces in MuJoCo are non-actuated forces that arise from mechanical element properties: joint springs, joint dampers, gravitational compensation springs, and tendon elasticity. These forces are applied automatically based on model parameters without requiring control input.
Context: Passive forces model the inherent mechanical properties of a system. Springs store and release energy based on displacement from an equilibrium position; dampers dissipate energy proportional to velocity. These forces contribute to the generalized force vector before the constraint solver runs.
Theoretical Basis
Passive forces follow classical mechanics principles:
- Spring forces: F = -k * (q - q_ref), a restoring force proportional to displacement from the reference position, storing potential energy U = 0.5 * k * (q - q_ref)^2
- Damping forces: F = -b * qdot, a dissipative force proportional to velocity, removing kinetic energy from the system
- Gravity compensation: Optional spring forces that counteract gravity at a reference configuration, enabling stable hovering or upright postures
- Tendon elasticity: Length-dependent forces in tendons that couple multiple joints, modeling transmission elements like cables and pulleys
Passive forces do not require energy input and are essential for modeling realistic mechanical systems with compliance and energy dissipation.
Related Pages
Implementations
Workflows
- (none yet)