Index
Parking Spot Detection and Mapping for Mobile Robots
Self-driving vehicles rely on detailed semantic maps of the environment for operating. In this work, we propose a method to autonomously generate such a semantic map enriched with knowledge of parking spot locations. Our method detects and uses parked vehicles in the surroundings to estimate parking lot topology and infer vacant parking spots via a graph-based approach. We show that our method works for parking lot structures in different environments, such as structured parking lots, unstructured/unmarked parking lots, and typical suburban environments. Using the proposed graph-based approach to infer the parking lot structure, we can extend the estimated parking spots by 57%, averaged over six different areas with ten trials each. We also show that the accuracy of our algorithm increases when combining multiple trials over multiple days. With ten trials combined, we managed to estimate the whole parking lot structure and detected all parking spots in four out of the six evaluated areas.
Processing pipeline of our method
For our parking spot detection and prediction method, different core technologies were combined. A coarse flowchart of our processing stream can be seen in the figure above. The algorithm can be divided into three different modules. The first module uses a simultaneous localization and mapping (SLAM) algorithm to generate a 3D map of the environment. The map is transformed into a 2D grid representation to compress the data size. The maps constructed from several trials over multiple days are merged into a single map in order to remove dynamic obstacles. The second module generates elevation maps from the LIDAR scan data and detects vehicles using a deep neural network (DNN). The module detects vehicles parked by other traffic participants in the surroundings and projects them onto the acquired map. Finally, the third module estimates the parking lot topology by employing a graph-based structure that combines the vehicle detections with the 2D grid map. Using the inferred topology, the method predicts vacant parking spots. Furthermore, the module tracks single parking spots over several trial to estimate the occupancy rate for each of the parking spots. This can help to estimate the probability of finding an unoccupied parking spot within an area, facilitating the task of finding a free parking spot.
Detailed overview
A detailed version, explaining the algorithm in detail can be found at
Parking Spot Detection and Mapping for Mobile Robots
Contact
Do you have any questions or comments? Please contact: