GATO

GPU-Accelerated Trajectory Optimization for Robotics.

---CURRENTLY UNDER CONSTRUCTION---

Directory structure:

Read notes on SQP solvers and source code structure in gato/.

• bindings/: Python bindings
• config/: Settings
  • cost_settings.cuh: cost function settings
  • sim_settings.cuh: mpc simulation settings
  • solver_settings.cuh: solver settings
• data/: Data (e.g. trajectories)
• dependencies/: External libraries and submodules
• docs/: Documentation
• dynamics/: Dynamics (plant files/grid files)
• examples/: Examples
• experiments/: Experiments for papers
• gato/: Library source
  • sim/: Simulation
  • solvers/: Trjactory optimization solvers
  • utils/: Utilities
  • gato.cuh: header file for gato toolbox
• tests/: Tests
• tools/: Tools

Settings:

Tweak settings in config/ and gato/gato.cuh

Python bindings:

See bindings to build and use Python bindings

Building:

git clone https://github.com/A2R-Lab/gato.git
cd gato
git submodule update --init --recursive
chmod +x tools/build.sh
chmod +x tools/cleanup.sh
chmod +x tools/bindings.sh
./tools/build.sh

Running:

Single mpc simulation:

./build/mpc

Batched sqp solve:

./build/multi-sqp

Track with iiwa14 (pcg):

./build/MPCGPU-pcg

Track with iiwa14 (qdldl):

./build/MPCGPU-qdldl