A comparative study of camera-lidar fusion algorithms for 3D object detection

Are you passionate about bridging the gap between advanced AI and robotics? Join us at the Lincoln Centre for Autonomous Systems (L-CAS) to work on an exciting project that combines state-of-the‑art robotic perception systems with real‑world robotics applications.
In this project, you are going to implement and compare different state-of-the‑art algorithms to detect objects (mainly humans) using a combination of Camera and LiDAR data. The algorithms chosen to be implemented will need to use a different methodology to fuse the data, including early, mid, or late fusion approaches.
You are expected to evaluate the performance of the algorithms implemented using the well‑known KITTI dataset for urban environments and a proprietary dataset for agricultural environments. The most important criteria to consider for the evaluation are the effect on detection performance when decreasing the quality of the data by a) decreasing the density of the point cloud data b) introducing distortions in the images, emulating fish eye camera distortions.

• Strong programming skills in Python
• Good understanding of computer vision
• Good understanding of Artificial Neural Networks
• Experience with ROS2
• Familiarity with Docker and Git Hub tools
• Excellent communication abilities

• Familiarity with 3D point clouds and depth images
• Familiarity with sensor fusion techniques

• Hands‑on experience with cutting‑edge AI technologies
• Integration into the dynamic L-CAS research team
• Possibility to participate in scientific writing and co-authoring a research paper

This is an internship position suitable for students pursuing a programme of study in computer science, robotics and/or AI at the University of Lincoln.