Pose Estimation for Off-road GNSS-denied Autonomous Navigation
Background
Eira Systems is a Swedish start-up developing electrically powered unmanned ground vehicles for logistics, CASEVAC and other supporting missions in demanding off-road environments. Eira has developed and tested the full-scale Eira-X prototype and has now received a contract to develop the framework for a second-generation platform together with a major Swedish defence actor. Eira is based at The Yard in Lindholmen, Gothenburg.
To enable unmanned operations, autonomy is needed. Eira is developing both driver-assist features that make remote-controlled operation more reliable and point-to-point autonomous navigation.
Problem
Navigation off-road is hard because every sensor has a weakness. GNSS can be blocked, jammed or spoofed. Lidar and radar emit signals that reveal the vehicle, and high-resolution versions are expensive. Wheel encoders suffer from slip. MEMS IMUs drift heavily, especially under vibration. Cameras fail in poor visibility. The sensors also differ in which degrees of freedom they observe well, how fast they measure, and over what time span they stay accurate. The challenge is therefore to fuse these complementary sensors into an accurate pose estimate, and to detect when one can no longer be trusted.
Goal description
The goal is to improve pose estimation and odometry in off-road terrain. The project has two directions, each with suggested subtopics: (A) visually limited conditions where normal cameras are not enough, and (B) conditions where cameras can see, but GNSS is unavailable and drift must be kept low.
A. Pose estimation in visually limited environments using passive sensors
- IMU-centric multi-modal sensor fusion & sensor degradation detection
- Learned inertial odometry
- Thermal-inertial odometry
- Privileged information learning
- Image dehazing & low-light image enhancement
B. GNSS-denied pose estimation with multiple non-overlapping stereo cameras
- Multi-modal sensor fusion & sensor degradation detection with additional sensors such as external IMUs, ground-facing radar and wheel encoders.
- Evaluating the pose estimation against RTK GNSS ground truth
- Front-to-rear camera loop closure
Our vehicle has stereo and thermal cameras, IMUs, wheel odometry, an Ouster lidar, ground-facing radar and an Nvidia Jetson AGX Orin. A 1/10-scale version has a similar sensor and compute setup.
Want to know more about how you could attack these topics? Reach out to us! We would also love to support you exploring your own ideas in a related topic. We also have a few more experimental topics if you want something even more challenging, and other thesis proposals in different areas.
Student background
We are looking for students who are passionate about autonomous ground vehicles, off-road driving, self-driving or physical AI in general. You only need to be ambitious and curious to apply. Applicants must, however, be citizens of a NATO or EU member state. We assess applications on the following, in order of priority (attach something for at least one of them):
- Projects (school or hobby) in a related area
- Personality and motivation to apply for this project
- Merits
- Knowledge (which bachelor's and master's, and relevant courses)
Apply before November 1st to andreas.munck@eirasystems.com