Robotics and autonomous systems · Case study
Autonomous exploration and semantic mapping robot
Our autonomous robot explores spaces it has never seen. It heads for the edge of its current map (frontier-based exploration), labels rooms and objects with computer vision as it builds the map (semantic fusion), and calculates the shortest route home, with no prior map and no human guidance.
The challenge
A robot dropped into an unknown building has to map it, understand what it sees, and still find its way home, all without a prior map or human help.
Our approach
We combined frontier-based exploration with semantic fusion: the robot keeps heading for the edge of what it has already mapped, labels the space with computer vision as it goes, and plans the shortest path back to its start.
What we built
- Frontier-based autonomous exploration of unknown terrain
- Camera-based mapping of rooms with semantic labels (semantic fusion)
- Shortest-path planning to return to the starting point
Highlights
- Runs without a prior map or human guidance
Technology
- Computer vision
- SLAM
- Path planning
- ROS
Client details are kept confidential. On a call we can walk you through this project in more depth, under NDA.