swarm-autonomy-simulation

This repository contains code for simulating the MCU autonomy related software. It includes C++ code to run simulations and python code to visualize results.

include and src

C++ code to run simulations.

• agent.h/cpp: simulates a complete swarm agent, each instance is given an ID number that is only used by the simulation environement when saving data.
• controller.h/cpp: mock controller used by the agents.

After a simulation is complete the data will be saved to the following file structure:

sim_data_<unix_time>
  <swarm_id>_states.csv
  <swarm_id>_desired_poses.csv
  <swarm_id>_trajectories
    <swarm_id>_<time>.csv
    <swarm_id>_<time2>.csv
  <next_swarm_id>_states.csv
  <next_swarm_id>_desired_poses.csv
  <next_swarm_id>_trajectories
    <next_swarm_id>_<time>.csv
    <next_swarm_id>_<time2>.csv
  ...

Where unix_time is the time the simulation started at, swarm_id is the ID number of the agent that generated that data and time/time2 is the time since simulation start the saved trajectory was generated at.

Build Instructions (Tested on Linux)

1. Go to package root.
2. run cmake -B build -G Ninja
3. Go to the new build directory
4. run ninja

sim_tools

Python code used to visulize simulation results.

• plot.py: code to plot trajectories

cpp_state_estimator

State estimator class for Pico 2 W with example .ino usage file.

py_state_estimator

Uses an old version of the vehicle model with thrust and sway propellors before we moved to the side-by-side thruster model.

• sensor_benchmarker.py: Computes the uncertainty of different sensor configurations for state estimation.
• sensor_specs.json: Specifications for various sensors used in the benchmarking.
• state_estimator.py: Simulates state estimation using an Extended Kalman Filter (EKF) and Rauch-Tung-Striebel (RTS) smoother.