Pioneer - UWaterloo's First Humanoid Robot

We are a team of undergraduate student from UWaterloo building a 160 cm humanoid robot. We are the first engineering design team in Canada building a humanoid!

Our humanoid robot will feature 2 arms + 2 legs and we are also developing a 22 DOF humanoid hand that can be modularly switchable with the arm gripper.

More info at Pioneer's website: https://watonomous.github.io/humanoid-docs/index.html

watod is our Docker Compose wrapper: it spins up one dev container per module (interfacing, perception, simulation, …), so you only run the stack your subteam needs.

cp watod-config.sh watod-config.local.sh   # set your ACTIVE_MODULES
./watod up -d                 # start your modules
./watod -t <service>          # shell into one (interfacing, perception, simulation_mj, ...)
./watod build <service>       # rebuild after a Dockerfile / dependency change
./watod down                  # stop

Your src/<module> folder is mounted into the container, so edits on your machine show up inside it immediately without rebuild.

watod-config.sh is shared defaults and is CI-guarded, so don't commit personal changes to it. Everything local goes in watod-config.local.sh (gitignored).

Isaac Lab sim (recommended): see docker/simulation/isaac_lab/README.md.

Two workflows share this repo:

• Real robot: embedded firmware ⇄ interfacing (CAN ⇄ ROS 2) ⇄ perception — driven live by teleop, or by a policy.
• Sim / learning: teleop collects demos in simulation scenes → robot_learning records datasets and trains policies, or simulation/humanoid_rl trains RL policies → deploy back through interfacing.

robot_learning vs RL: src/robot_learning learns policies from recorded datasets (sim or real), e.g. ACT, SmolVLA, pi0.5. RL is sim-based (Isaac Lab), so it lives in src/simulation/humanoid_rl*.

humanoid
├── watod, watod-config.sh   # container orchestrator + module config
├── modules/  docker/        # compose files + Dockerfiles, per module
├── src/
│   ├── pioneer_humanoid/    # the robot definition (imported everywhere)
│   ├── interfacing/         # CAN ⇄ ROS 2 bridge, real-arm bring-up
│   ├── embedded/            # STM32 / ESP32S3 motor firmware
│   ├── perception/          # cameras, voxel grid
│   ├── simulation/          # Isaac Lab + MuJoCo: RL tasks, teleop scenes
│   ├── teleop/              # drive the arm (sim or real)
│   ├── robot_learning/      # record demos, train / eval policies
│   └── common_msgs/         # shared ROS 2 messages
├── assets/                  # robot URDF / USD / meshes, scene props
├── models/                  # frozen RL policy baselines
├── outputs/                 # training runs (gitignored)
└── docs/                    # pointer to the docs site

Full detail — src/simulation/README.md. Other areas: each has its own README.md.

Connect → calibrate → visualize → move, top to bottom: src/interfacing/README.md

./src/interfacing/can/scripts/can_udev.sh install   # once per host → /dev/canable

• CI does not build simulation_* or embedded. A green check on those means nothing was built, so test them yourself.
• Set a unique ROS_DOMAIN_ID (0–232) in watod-config.local.sh if someone else runs ROS on the same subnet, or you will see each other's topics.