The major topics of research interest of the MIR Lab include (but not limited to):
We specifically focus on driving control based on behavior cloning using deep learning techniques.
Transforming a Model Car to a Self-Driving Car
Simulation of Self-Driving Car on TORCS
We collected a human driver’s behavior in a racing car simulator. The driver’s view image data with steering angles and throttle values were collected for approximately four hours. Then we trained a Convolutional Neural Network model and had it to drive a racing car. Here is a fully autonomous test drive. Source code can be found at https://github.com/jrkwon/mir_torcs.
We used the same CNN model on a different track to show the flexibility of this approach. The followings show as successful driving on a different track from one where the model was trained.
ROS Gazebo and Rviz with Chevy Bolt
Drive by Wire Testing
Autoware is an open-source software for urban autonomous driving. It supports followings: 3D Localization, 3D Mapping, Path Planning, Path Following, Accel/Brake/Steering Control, Data Logging, Car/Pedestrian/Object Detection, Traffic Signal Detection, Traffic Light Recognition, Lane Detection, Object Tracking, Sensor Calibration, Sensor Fusion, Cloud-oriented Maps, Connected Automation, Smartphone Navigation, Software Simulation, Virtual Reality.
This means that Autoware can be a good reference for autonomous driving related research.
Gazebo Demo of dbw_mkz
This example is from Dataspeed that developed an ADAS kit for Lincoln MKZ. The demo below shows an example of lane detection in the Gazebo environment.
Projects from Udacity Self-Driving Car Nano Degree Term1
Intelligent Mobile Robotics and Applications:
- Architectures, localization, mapping, exploration, navigation
- Multi-robot systems, coordination, coverage control, task allocation
- Applications: domestic service applications, commercial, health-care
Wireless sensor networks and applications:
- Architectures, sensor localizations, communication topologies, coverage
- Mobile sensor networks, coverage control, workload distribution
- Energy harvesting strategies
- Applications: security & surveillance, environmental monitoring, personal health monitoring
- Evolutionary computation: performance enhancement & hardware acceleration
- Machine learning / Neural networks / Neuroevolution / Neuroinformatics
- Application of recent bio-inspired strategies for problem solving