Text2Pos: Text-to-Point-Cloud Cross-Modal Localization
It addresses the need for natural language-based communication in robotics, enabling tasks like vehicle pick-up or goods delivery, but is incremental as it builds on existing datasets and methods.
The paper tackles the problem of cross-modal text-to-point-cloud localization for mobile robots, proposing Text2Pos to align textual descriptions with point cloud cues, achieving 65% of queries localized within 15m for top-10 retrievals on the new KITTI360Pose dataset.
Natural language-based communication with mobile devices and home appliances is becoming increasingly popular and has the potential to become natural for communicating with mobile robots in the future. Towards this goal, we investigate cross-modal text-to-point-cloud localization that will allow us to specify, for example, a vehicle pick-up or goods delivery location. In particular, we propose Text2Pos, a cross-modal localization module that learns to align textual descriptions with localization cues in a coarse- to-fine manner. Given a point cloud of the environment, Text2Pos locates a position that is specified via a natural language-based description of the immediate surroundings. To train Text2Pos and study its performance, we construct KITTI360Pose, the first dataset for this task based on the recently introduced KITTI360 dataset. Our experiments show that we can localize 65% of textual queries within 15m distance to query locations for top-10 retrieved locations. This is a starting point that we hope will spark future developments towards language-based navigation.