2027 Internship State Estimation, Learned Mapping & Semantic SLAM
Core
Building learned mapping and localization systems for autonomous excavators to navigate changing construction terrain.
Role type
Intern, State Estimation & Learned Mapping
Builds
3D semantic maps and localization pipelines for heavy construction equipment
Domain
Construction robotics / Autonomous systems
Deliverable
production ML models
Required skills
Python, PyTorch, 3D geometry, SLAM fundamentals, point cloud processing, sensor data handling
Preferred skills
Learned SLAM, neural scene representations (NeRF, 3D Gaussian splatting), vision foundation models, lidar processing, multi-sensor fusion, ROS
Technologies
PyTorch, ROS, Lidar, Cameras
Responsibilities
Prototype learned SLAM and mapping methods, fuse lidar and camera segmentation into 3D semantic maps, develop methods for changing terrain and occlusion, train models on fleet data, collaborate with perception and planning teams, deliver documented prototypes and experimental results
