Skip to main content

Boston Dynamics Spot

by Boston Dynamics

Boston Dynamics Spot

Boston Dynamics Spot is an agile mobile robot that navigates terrain with unprecedented mobility, allowing you to automate routine inspection tasks and data capture safely, accurately, and frequently. This high-fidelity MuJoCo simulation includes the full quadruped body with optional robotic arm attachment. Features 12 degrees of freedom for the base quadruped, enabling dynamic gaits, stair climbing, and complex terrain navigation. The model includes detailed collision geometry and visual meshes for realistic physics simulation.

Specifications

Manufacturer
Boston Dynamics
Type
Quadruped Robot
Environment
Simulated
Category
MuJoCo Physics Simulator
Controller
MuJoCo
Sensors
Joint position
Joint velocity
Joint torque
Body orientation
Body velocity
Ground contact
Foot pressure
Actuators
Position servo
Velocity servo
Torque control
Connectivity
ZMQ
Local

Getting started

Boston Dynamics Spot - FEAGI Integration Guide

Overview

The Boston Dynamics Spot simulation provides a high-fidelity quadruped robot platform from MuJoCo Menagerie. With 12 degrees of freedom and advanced kinematics, Spot challenges FEAGI to learn complex quadruped locomotion, balance control, and dynamic gait generation across varied terrains.

Model Specifications

  • Degrees of Freedom: 12 (3 per leg: hip x/y, knee)
  • Actuators: 12 position-controlled servos
  • Control Frequency: 50 Hz
  • Simulation Timestep: 0.002s
  • Total Mass: 32.5 kg
  • Mesh Files: 54 detailed collision and visual geometries

Sensory Capabilities

Joint Sensors

  • Joint Position: Current angle of each actuated joint (12 values)
  • Joint Velocity: Angular velocity of each joint (12 values)
  • Joint Torque: Measured torque at each joint (12 values)

Body State Sensors

  • Body Orientation: Quaternion representing torso rotation in 3D space
  • Body Velocity: Linear and angular velocity of the torso
  • Body Acceleration: Linear and angular acceleration

Contact Sensors

  • Ground Contact: Binary sensors for each foot (4 values)
  • Foot Pressure: Contact force magnitude at each foot

Motor Capabilities

Actuator Types

  1. Position Control: Target joint angles (default mode)
  2. Velocity Control: Target joint velocities
  3. Torque Control: Direct torque commands

Joint Organization

  • Front-Left Leg: Hip-X, Hip-Y, Knee
  • Front-Right Leg: Hip-X, Hip-Y, Knee
  • Rear-Left Leg: Hip-X, Hip-Y, Knee
  • Rear-Right Leg: Hip-X, Hip-Y, Knee

Joint Ranges

  • Hip-X (abduction/adduction): ±45°
  • Hip-Y (flexion/extension): -60° to +90°
  • Knee: -150° to -40°

FEAGI Integration

Cortical Mapping

Input (Sensory):
  • i_joint_position: 12 neurons
  • i_joint_velocity: 12 neurons
  • i_joint_torque: 12 neurons
  • i_body_orientation: 4 neurons (quaternion)
  • i_body_velocity: 6 neurons (linear + angular)
  • i_ground_contact: 4 neurons (one per foot)
Output (Motor):
  • o_joint_control: 12 neurons (position/velocity/torque commands)

Neural Circuit Design Tips

  1. Central Pattern Generators (CPGs): Create oscillatory circuits for rhythmic leg movements
  2. Diagonal Coordination: Coordinate diagonal leg pairs (FL-RR, FR-RL) for trot gait
  3. Phase Relationships: Manage phase offsets between legs for smooth gaits
  4. Balance Control: Use body orientation feedback to maintain upright posture
  5. Gait Transitions: Switch between walk, trot, and bound gaits based on speed
  6. Terrain Adaptation: Adjust leg lift height based on ground contact feedback

Beginner

  1. Stand Still: Maintain upright posture with all feet on ground
  2. Single Leg Lift: Lift and lower one leg while maintaining balance
  3. Weight Shifting: Transfer weight between diagonal pairs

Intermediate

  1. Walking Gait: Achieve slow forward locomotion with walking gait
  2. Trotting: Implement diagonal pair coordination for faster movement
  3. Turning: Execute controlled turns while maintaining gait
  4. Backwards Movement: Reverse locomotion

Advanced

  1. Stair Climbing: Navigate stairs and elevated surfaces
  2. Bounding: High-speed galloping gait with all legs off ground
  3. Uneven Terrain: Adapt gait to slopes, obstacles, and rough surfaces
  4. Energy Efficiency: Minimize torque consumption while maintaining speed
  5. Recovery Behaviors: React to external disturbances and regain balance

Model Entry Points

The Spot model provides multiple entry points:

  • scene.xml (default): Complete environment with floor, lighting, and skybox
  • scene_arm.xml: Spot with robotic arm for manipulation tasks
  • spot.xml: Base robot without environment (for custom scenes)

Troubleshooting

Robot Falls Over

  • Check body orientation feedback is connected
  • Verify ground contact sensors provide feedback
  • Ensure joint angles remain within valid ranges
  • Increase initial torso height

Poor Locomotion

  • Verify diagonal leg coordination (FL-RR, FR-RL)
  • Check CPG frequency matches intended gait speed
  • Ensure ground contact triggers stance/swing transitions
  • Review joint torque limits

High Computational Load

  • Reduce visual mesh resolution if needed
  • Use headless mode (no GUI) for faster simulation
  • Lower control frequency if CPU-limited
  • Consider smaller time steps for stability

Resources

License

This model is provided under the BSD-3-Clause license by Boston Dynamics and Google DeepMind. See the LICENSE file in the model directory for full terms.

Support

For integration questions or issues:

Connecting to FEAGI

Boston Dynamics Spot - Connectivity

Connection Type

Local WebSocket + ZMQ

The Boston Dynamics Spot simulation runs on your local machine and connects to FEAGI via:

  • ZMQ for sensory data (joint positions, velocities, body state)
  • ZMQ for motor commands (joint control signals)
  • HTTP for registration and metadata

Network Requirements

Ports Used

  • 5555: Sensory data (ZMQ PUB socket)
  • 30005: Motor commands (ZMQ SUB socket)
  • 30001: Agent registration (TCP)

Firewall Configuration

Ensure these ports are open for localhost communication.

Controller Setup

Installation

The generic MuJoCo controller is installed once and can run any MuJoCo model.

Via FEAGI Desktop:
  1. Navigate to Embodiment Explorer
  2. Select "MuJoCo Physics Simulator" category
  3. Click on "Boston Dynamics Spot"
  4. Click "Build Your Own" to launch

Launch Command

The controller automatically handles model loading with the correct entry point (scene.xml).

Model Variants

This embodiment includes multiple model configurations:

  • scene.xml: Full environment with floor and skybox (default entry point)
  • scene_arm.xml: Spot with robotic arm attachment
  • spot.xml: Base robot model only

Performance Notes

This model includes 54 detailed mesh files for high-fidelity visualization. Performance considerations:

  • Runs smoothly on modern hardware (2018+ Mac/PC)
  • GPU acceleration recommended for visual rendering
  • Headless mode available for faster physics-only simulation

Support

For connectivity issues:

View marketplace for this embodiment · Back to embodiment explorer