Unstructured Navigation Platform

Overview

The Unstructured Navigation Platform is designed to facilitate autonomous navigation in unstructured environments. It leverages ROS (Robot Operating System) on NVIDIA Jetson platforms to perform feature extraction, path planning, and control, allowing robots to navigate complex terrains.

Introduction
System Requirements
Getting Started
Low-Level Control and Sensor Nodes
Training the Steago Model
Troubleshooting
Contributing
Contact

Introduction

This platform is built to enable autonomous navigation in environments that are typically challenging for robots to traverse. It includes feature extraction for understanding the environment, and path planning for safe and efficient navigation.

System Requirements

Hardware:
- NVIDIA Jetson AGX Orin
- LiDAR sensor (Livox)
- ZED Stereo Camera
Software:
- Ubuntu 20.04
- Docker CE with NVIDIA runtime
- ROS Noetic
- Git

Getting Started

1. Clone the Repositories into workspace

git clone https://github.com/amilearning/unstruct_navigation.git -b orin2
git clone https://github.com/amilearning/unsup_seg_pkg.git -b orin
git clone https://github.com/amilearning/Hama_nav.git

Baseline Repository: Self-Supervised Segmentation " https://github.com/leggedrobotics/self_supervised_segmentation "

2. Build the ROS Packages

After cloning the repositories, build the ROS workspace:

cd ~/workspace/
catkin_make -DCMAKE_BUILD_TYPE=Release
source devel/setup.bash

3. Run the Docker Container

Use the provided shell script torch_docker2.sh to start the Docker container:

chmod +x torch_docker2.sh
./torch_docker2.sh

Alternatively, you can run the Docker container manually (steogo_ws is the workspace directory in local env)

docker run --runtime nvidia -it --rm --network host \
  --volume /tmp/argus_socket:/tmp/argus_socket \
  --volume /etc/enctune.conf:/etc/enctune.conf \
  --volume /etc/nv_tegra_release:/etc/nv_tegra_release \
  --volume /tmp/nv_jetson_model:/tmp/nv_jetson_model \
  --volume /var/run/dbus:/var/run/dbus \
  --volume /var/run/avahi-daemon/socket:/var/run/avahi-daemon/socket \
  --volume /var/run/docker.sock:/var/run/docker.sock \
  --volume /home/hmcl/jetson-containers/data:/data \
  --device /dev/snd \
  --device /dev/bus/usb \
  --device /dev/video0 \
  --device /dev/video1 \
  --device /dev/i2c-0 --device /dev/i2c-1 --device /dev/i2c-2 \
  --device /dev/i2c-3 --device /dev/i2c-4 --device /dev/i2c-5 \
  --device /dev/i2c-6 --device /dev/i2c-7 --device /dev/i2c-8 \
  --device /dev/i2c-9 \
  -v /run/jtop.sock:/run/jtop.sock \
  -v /home/hmcl/steogo_ws:/home/stego_ws \
  amilearning/unstruct_nav_l4t:v1

4. Launch the ROS Nodes

Start the Feature Extraction Node:

roslaunch unstruct_navigation_ros unstruct_visual_navigation.launch

Start the Path Planning Node:

roslaunch unstruct_navigation_ros unstruct_planner.launch

Low-Level Control and Sensor Nodes

The Hama_nav repository contains essential nodes for low-level control and additional sensor integration.

Repository URL: Hama_nav

Low-Level Control Initialization:
```
roslaunch lowevel_ctrl low_level.launch
```
This command initializes the low-level controllers responsible for direct hardware interactions, such as motor control and sensor data processing.
LiDAR, and state estimator initialization: check alias in bashrc..
MAVROS Initialization:
```
rosrun mavros px4.launch 
```
MAVROS is used to interface ROS with the robot's autopilot system, allowing for telemetry and control data exchange.

PX4 Firmware

The PX4 firmware is designed for the Pixhawk flight controller and is essential for converting signals to control servo and motor PWM outputs. This firmware allows for precise control of actuators in robotic platforms, particularly for off-road and autonomous navigation applications.

Repository URL: offroad_px4autopilot

This repository contains the custom PX4 firmware tailored for off-road autonomous vehicles. The firmware is adapted to manage the specific requirements of converting control signals to PWM outputs for servos and motors.

Installation and Setup

Clone the PX4 Firmware Repository:

git clone https://github.com/amilearning/offroad_px4autopilot.git ~/px4_firmware

Training the Steago Model

To train the Steago model, navigate to the unsup_seg_pkg directory and run the training script:

Troubleshooting

Docker Issues: Ensure the NVIDIA runtime is correctly installed and that all volumes and devices are properly mapped.
ROS Nodes Not Communicating: Verify that the ROS environment is correctly sourced and that all nodes are runnin (even inside of docker).

amilearning/Hama_nav