Megvii’s robot localization and navigation technology provide autonomous flexible path planning together with highly effective obstacle avoidance. It is achieved by fusing information from multiple sensors such as visual and laser to reconstruct 3D environmental map in real-time.
Simultaneous Localization and Mapping (SLAM)
a. By combining vision, laser and other sensors, our robot filters moving objects and is well-adapted in dynamic environments, and achieves localization precision of ±1cm and position repeatability of ±5mm.
b. Our robot can start and reboot anywhere anytime without interference with scene recognition. By coordinating with other agents, it automatically evaluates environmental changes to produce reliable map update.
Related Products
Navigation
a. With high-precision maps and obstacle avoidance system, our navigation is agile, safe, and efficient.
b. Our robot identifies various obstacles including human, potholes, hovering objects and etc. Equipped with different avoidance strategy, it achieves extremely high safety standards.
Related Products
Robotic Grasping
a.Grasp point and suck point selection. Accurately finding parts of the object that the robot can easily grasp based on RGBD information. Applicable to unseen objects and environments.
b.Instance segmentation. Parsing an image to individual objects. Compared to existing solutions, our algorithm more robustly handles crowded cases.
c.Box detection. Finding the positions of boxes so that the robot can manipulate them. Especially useful in warehouse settings.
Related Products
3D Perception
6D pose estimation. Calculating 6D (XYZ + rotation) pose of objects with SOTA accuracy on public benchmarks and great robustness to illumination and occlusion.
Related Products
