Friday Sneak Peek: Unveiling 📡Radar-Based SLAM 🗺️
💥 In a previous post we have shown our our advancements with LiDAR based localization and mapping, but today, we’re excited to reveal something new: 📡Radar-Based SLAM 🗺️ (Simultaneous Localization and Mapping).
Take a look at this fascinating video 🎥 created by our talented colleague, Marco de Böck.
📍Driving through the always-busy campus of Delft University of Technology, notice how our innovative radar-only method reliably maps the environment over time, and estimates the car's position and velocity correctly despite the numerous moving objects (so many cyclists in the Netherlands 🇳🇱🚴♀️!).
Remember, only radar is used to do this, no camera, no gps, no lidar, not even IMU.
Why is radar-only ego-pose and velocity estimation is so important?
🔹 Increased Redundancy for Localization: It serves as an additional input for localization, crucial in GPS-denied areas like parking garages or in adverse weather.
🔹 Ego-Motion Compensation: Radar uniquely measures the velocity of objects relative to the vehicle. However, how to tell if the ego-car is moving, or the cyclist ahead of us? To do this, we need to know our own velocity first. By doing radar SLAM, we achieve that without depending on other sensors, making our system independent.
🔹 Enhanced Scene Awareness: A detailed map of both moving and static objects provides valuable data to our AI-driven, scene-aware radar perception stack, enhancing performance and safety.
Stay tuned as we continue to extend the capabilities of radar technology at Perciv AI, and congrats to Marco de Böck for the nice work!
#ai4radar #perceptionradar #perciv #SLAM #radar_only
#RadarTechnology #AI #AutonomousDriving #Innovation #PercivAI #TUDelft
Andras Palffy Srimannarayana Baratam Balazs Szekeres
Distinguished Professor Emeritus, Entrepreneur, National Academy of Engineering
1moGreat work SILC team!