Project
AgiBot World
Manipulation
Manipulation- Cutting-edge Sensor and Hardware Design.
- Wide-spectrum of Scenario Coverage.
- Quality Assurance with Human-in-the-loop.
OpenDV
GenerationWorld Model
GenerationWorld Model- The largest driving video dataset to date, containing more than 1700 hours of real-world driving videos.
DriveLM
LanguagePlanning
LanguagePlanning- Driving with Graph Visual Question Answering.
- Facilitating the Perception, Prediction, Planning, Behavior, Motion tasks with human-written reasoning logic as a connection in between.
- Autonomous Grand Challenge @ CVPR 2024 with 152 teams from 14 countries and regions participating.
OpenScene
Occupancy
Occupancy- Covering a wide span of over 120 hours and providing occupancy labels collected in various cities.
- Autonomous Grand Challenge @ CVPR 2024 with 212 teams from 13 countries and regions participating.
OpenLane-V2
Topology Recognition3D Lane DetectionTraffic Element Recognition
Topology Recognition3D Lane DetectionTraffic Element Recognition- The world's first perception and reasoning benchmark for scene structure in autonomous driving.
- Autonomous Grand Challenge @ CVPR 2024 with 120 teams from 10 countries and regions participating.
- Autonomous Driving Challenge @ CVPR 2023 with 34 teams from 4 countries and regions participating.
OpenLane
3D Lane Detection
3D Lane Detection- The first real-world and the largest scaled 3D lane dataset to date.
- Owning 200K frames and over 880K lanes.