Skip to main content

​
Overview

The control utilities module provides the actuator mapping and helper functions to apply inverse kinematics solutions to the MuJoCo simulation. It handles the leg-specific transformations, sign corrections, and offset adjustments required to map joint angles to actuator control slots.

​
apply_leg_angles()

Maps IK output angles into the correct actuator ordering for a specified leg. 
def apply_leg_angles(
    ctrl: mujoco.MjData,
    leg: str,
    angles: Tuple[float, float, float]
) -> None

​
Parameters

​
ctrl
mujoco.MjData
required
MuJoCo data object. The function writes target actuator positions to ctrl.ctrl.
​
leg
str
required
Leg identifier. Must be one of:
  • "FL" - Front Left
  • "FR" - Front Right
  • "RL" - Rear Left
  • "RR" - Rear Right
​
angles
Tuple[float, float, float]
required
Joint angles (tilt, shoulder_left, shoulder_right) in radians from IK solver.

​
Returns

None: Modifies ctrl.ctrl in place.

​
Usage Example

import mujoco
from ik import solve_leg_ik_3dof
from utils.control_utils import apply_leg_angles
import numpy as np

# Load MuJoCo model
model = mujoco.MjModel.from_xml_path("robot.xml")
data = mujoco.MjData(model)

# Define target position for front-left leg
target = np.array([0.0, 0.01, -0.04])  # 1cm forward, 4cm down

# Solve IK
angles = solve_leg_ik_3dof(target)

if angles:
    # Apply angles to actuators
    apply_leg_angles(data, "FL", angles)
    
    # Step simulation
    mujoco.mj_step(model, data)

​
Multi-Leg Control Example

# Control all four legs simultaneously
from utils.control_utils import apply_leg_angles
import numpy as np

# Define targets for each leg
targets = {
    "FL": np.array([0.0, 0.01, -0.04]),
    "FR": np.array([0.0, 0.01, -0.04]),
    "RL": np.array([0.0, -0.01, -0.04]),
    "RR": np.array([0.0, -0.01, -0.04]),
}

# Solve IK and apply for each leg
for leg_name, target in targets.items():
    angles = solve_leg_ik_3dof(target)
    if angles:
        apply_leg_angles(data, leg_name, angles)

# Execute simulation step
mujoco.mj_step(model, data)

​
LEG_CONTROL

Actuator mapping configuration dictionary containing the actuator indices, sign corrections, and offset adjustments for each leg. 
LEG_CONTROL: Dict[str, LegControl] = {
    "FL": LegControl(indices=(0, 1, 2), sign=-1.0, offset=-np.pi),
    "FR": LegControl(indices=(6, 7, 8), sign=1.0, offset=np.pi),
    "RL": LegControl(indices=(3, 4, 5), sign=-1.0, offset=-np.pi),
    "RR": LegControl(indices=(9, 10, 11), sign=1.0, offset=np.pi),
}

​
LegControl Dataclass

@dataclass(frozen=True)
class LegControl:
    indices: Tuple[int, int, int]  # (left_shoulder, right_shoulder, tilt)
    sign: float                     # Sign correction for leg orientation
    offset: float                   # Angular offset for right shoulder

​
Actuator Index Mapping

The actuator control array (ctrl.ctrl) has 12 slots corresponding to the joints:
IndexJointLeg
0Left ShoulderFL
1Right ShoulderFL
2TiltFL
3Left ShoulderRL
4Right ShoulderRL
5TiltRL
6Left ShoulderFR
7Right ShoulderFR
8TiltFR
9Left ShoulderRR
10Right ShoulderRR
11TiltRR

​
Sign and Offset Corrections

  • Left legs (FL, RL): sign = -1.0, offset = -π
  • Right legs (FR, RR): sign = 1.0, offset = π
These corrections account for:
  1. Mirror symmetry between left and right legs
  2. Physical mounting orientation differences
  3. URDF/XML joint zero position conventions

​
Usage Notes

Always call apply_leg_angles() after solving IK and before stepping the simulation. The function handles all necessary coordinate transformations automatically.
Ensure the leg identifier string matches exactly: "FL", "FR", "RL", or "RR". Invalid leg names will raise a KeyError.

Build docs developers (and LLMs) love

Get started for freeTalk to us