Obstacle Avoidance Isaac Sim

Mobile robot equipped with lidar for exploration of environment. This directory contains simulations, models to train the robot.

Description

This simulation is used to create custom tracks and drive around with robot to collect new data. We collect 360 degree scan of the track, x_axis and y_axis of the left stick. This simulation supports regression approach to the problem of autonomous navigation

Steering	Wheel velocities
	$$V_L = v - \frac{\omega \cdot L}{2 \cdot r}$$ $$V_R = v + \frac{\omega \cdot L}{2 \cdot r}$$ $V_L$ - velocity of the left side; $V_R$ - velocity of the right side; $v$ - linear velocity; $\omega$ - angular velocity; $L$ - distance between wheels; $r$ - radius of the wheel.
Velocities are adjusted based on the position of the lef stick on the X, Y axises.	Linear velocity is a constant value, that is set in the program. Angular velocity is compute by multiplication of max angular velocity and predicted value by the neural network.

Installation & Use

Follow this instruction:

1. Download isaac sim from this website
2. In order to launch issac sim run following command on (windows):

cd C:\isaacsim
isaac-sim.bat

3. In the issac sim window open project located in simulations directory

Dataset

The dataset has been shared on the kaggle website here. It was used to train the robot. If you want to improve overall performance, please drive robot on the new tracks. The more angles in training, the more robust the robot performance is.

Contribution

Student Information	Image
Michał Zientek is the primary author and creator of the ObstacleAvoidanceIsaacSim project. He was responsible for all stages of development, including: - Creating simulation - Collecting data - Training of neural networks - Testing the performance of the robot - Producing documentation

License

This project is licensed under the MIT License. See the LICENSE file for details.