Compared to roadways with structured characteristics, sidewalk scenarios in urban environments are often more complex and variable, and are also easily encountered with special and unknown challenges. The Continental-NTU Delivery Robotic Platform dataset was constructed to provide a class of multimodal sensor information sets that can be used for sidewalk navigation in unstructured environments, and to provide validation support for navigation and decision-making algorithms of last-mile delivery robots. The sensor information in the released dataset not only covers 3D LiDAR, camera, and inertial measurement unit data for sidewalk navigation, but also provides ground truth odometry and mesh data for verification and computer simulation, respectively. This dataset was collected in real-time by the prototype food delivery robot platform developed by Continental and organized in JSON file format. In addition, eight routes that can be used for food delivery tasks were selected for recording in the campus environment of Nanyang Technological University. In this paper, the composition of this dataset is described in detail. Furthermore, a variety of existing SLAM algorithms are used on the basis of this dataset for validating reliability and providing benchmarks. The website for access: https://ntu-conti-a2.github.io/Continental-NTU-Dataset.
Continental-NTU Delivery Robot Dataset for Perception and Navigation on Footpath
2023-09-24
7612612 byte
Conference paper
Electronic Resource
English
Footpath cycling trials in Victoria
British Library Conference Proceedings | 1992
|British Library Conference Proceedings | 1992
|Weed control cuts footpath maintenance costs
British Library Online Contents | 1998
Nunawading footpath cycling trial: a case study
British Library Conference Proceedings | 1992
|Novel barrel steel structure spiral footpath bridge with elevators
European Patent Office | 2025
|