JD.com Delivery Robots: Advanced Localization, Sensor Fusion, and AI‑Driven Navigation
The article details JD.com’s 3.5‑generation delivery robots, explaining their high‑precision multi‑sensor localization, deep‑learning perception, reinforcement‑learning control, extensive patent portfolio, and future challenges, while also inviting readers to vote for the robots in a national patent competition.
JD.com’s delivery robots, now in their 3.5‑generation, have achieved a global first‑time full‑scenario regular operation, covering residential, campus, and industrial park deliveries.
Where Am I
The robots rely on high‑precision positioning using multi‑sensor fusion of inertial navigation boards, LiDAR point‑cloud matching, visual odometry, and wheel encoders, achieving centimeter‑level accuracy.
How Do I Move
To navigate complex environments, the system processes massive real‑world data with deep‑learning‑based point‑cloud clustering and robust tracking, maintains 50 Hz prediction, and builds real‑time local maps via SLAM, integrating radar and visual information.
Intelligent Control
Deep learning and reinforcement learning are applied to motion control, enabling the robot to recognize traffic signs, lane markings, and adapt to varying lighting and weather, ensuring safe and smooth operation across speed ranges.
Patents and Future
Hundreds of patents protect the robot’s chassis, electrical, perception, algorithms, and safety technologies, while future challenges include scaling high‑precision maps and reducing LiDAR costs to expand autonomous delivery nationwide.
Readers are invited to support the robot’s “China Good Patent” voting campaign through various online channels.
JD Retail Technology
Official platform of JD Retail Technology, delivering insightful R&D news and a deep look into the lives and work of technologists.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.