The Embodied Zoo

Vetted Specifications for Embodied AI Platforms, Cyber-Physical Machines, and Robotics Archetypes

The Embodied Zoo is the authoritative registry for cyber-physical machine embodiments in mlsysim. Every platform is strictly typed (pint.Quantity), provenance-tracked against manufacturer datasheets, and categorized under the four canonical Physical AI machine archetypes.

TipHow to use this page

Reference these platforms when reasoning about physical constraints: stopping distances, contact impedances, actuator saturation, control loop timing, and onboard compute power. Load any platform directly in Python: platform = mlsysim.Embodied.Quadruped.Spot.

Cyber-Physical Embodiments

Platform Archetype Mass Payload Max Speed Control Rate Onboard Compute Transmission
Boston Dynamics Spot Class 1: Mobility 32.7 kg 14.0 kg 1.6 m/s 1,000.0 Hz NVIDIA Jetson AGX Orin Quasi-Direct Drive (QDD) / Brushless Outrunner
Boston Dynamics Atlas Class 4: Humanoid Capstone 89.0 kg 11.0 kg 2.5 m/s 1,000.0 Hz NVIDIA Jetson AGX Orin High-bandwidth electro-hydraulic / QDD electric
Unitree H1 Class 4: Humanoid Capstone 47.0 kg 30.0 kg 3.3 m/s 1,000.0 Hz NVIDIA Jetson AGX Orin High-torque M107 joint motors (360 N·m)
ALOHA Bimanual Manipulator Class 2: Manipulation 15.0 kg 0.5 kg 1.0 m/s 50.0 Hz NVIDIA Jetson AGX Orin Dynamixel servo actuators / 4-bar linkage gripper
Franka Emika Panda Class 2: Manipulation 18.0 kg 3.0 kg 2.0 m/s 1,000.0 Hz NVIDIA Jetson AGX Orin Harmonic drive with integrated joint torque sensors
DJI Matrice 350 RTK Class 1: Aerial Mobility 6.5 kg 2.7 kg 23.0 m/s 400.0 Hz NVIDIA Jetson Orin Nano Direct-drive brushless DC motors (FOC)
Heavy Logistics AMR Class 1: Industrial Mobility 250.0 kg 800.0 kg 1.8 m/s 100.0 Hz NVIDIA Jetson AGX Orin Differential drive with planetary reduction
Industrial Logistics AMR Class 1: Industrial Mobility 150.0 kg 500.0 kg 1.8 m/s 100.0 Hz NVIDIA Jetson AGX Orin Differential wheel drive with planetary gearboxes
Warehouse Fulfillment AMR Class 1: Industrial Mobility 300.0 kg 1,000.0 kg 1.5 m/s 100.0 Hz NVIDIA Jetson AGX Orin Differential dual-drive in-wheel brushless motors
Autonomous Vehicle / Robotaxi Class 1: Heavy Mobility 2,200.0 kg — 33.3 m/s 100.0 Hz NVIDIA DRIVE Thor Electric multi-motor powertrain
Uber ATG Volvo XC90 Class 1: Heavy Mobility 2,100.0 kg — 25.0 m/s 100.0 Hz NVIDIA Jetson AGX Orin Hydraulic friction braking with steer-by-wire

The Four Canonical Machine Archetypes

Physical AI embodiments are governed by the conservation law that binds their actuators first:

  1. Class 1: Mass-Dominated Mobility Systems (Spot, Drones, AMRs, Autonomous Vehicles)
    • Bound first by momentum conservation, kinetic energy dissipation (\(E_k = \frac{1}{2}m v^2\)), and spatial clearance (\(d_{\text{stop}} = v \cdot \tau + \frac{v^2}{2a_{\max}}\)).
  2. Class 2: Contact-Dominated Manipulation Systems (Franka Emika Panda, articulated arms)
    • Bound first by mechanical contact impedance, joint torque saturation, and structural yield stress (\(F = k \Delta x\)).
  3. Class 3: Flow-Dominated Process & Energy Systems (Thermal plants, battery management containers)
    • Bound first by lumped thermal capacity, transport delay, and chemical phase transitions.
  4. Class 4: Integrative Humanoid Capstones (Atlas, Unitree H1)
    • Integrates all three physical classes simultaneously on a single dynamic chassis.

Python Access

import mlsysim

# Load platforms
spot = mlsysim.Embodied.Quadruped.Spot
atlas = mlsysim.Embodied.Humanoid.Atlas
panda = mlsysim.Embodied.Manipulator.Panda
amr = mlsysim.Embodied.AMR.LogisticsAMR
robotaxi = mlsysim.Embodied.Vehicle.Robotaxi

# Access vetted physical properties
print(spot.mass)               # 32.7 kg
print(amr.max_velocity)        # 1.8 m/s
print(robotaxi.compute_soc)    # NVIDIA DRIVE Thor
Back to top