RoboSense and BitRobot Form Strategic Partnership, Release 100-Hour 3D Work Dataset

leaderobot.com reported that RoboSense and BitRobot announced a strategic partnership on Nov. 11 during CoRL 2026. One outcome, the report said, was the release of “RoboCap LiDAR-100,” a dataset containing 100 hours of real-world work in kitchens, auto repair shops and automotive detailing settings. According to the report, the dataset synchronizes LiDAR, RGB images and inertial measurement data, and outputs depth data at QVGA resolution.

The leaderobot.com report points to efforts to prepare work data that synchronizes LiDAR, RGB images and inertial measurement for use in robot training. The release of a dataset alone does not establish improved real-world work capabilities. The accuracy of spatial measurements and their correlation with real-robot evaluations will be key to assessing the dataset’s value as a training resource.

The reported achievement centers on collecting data that records real-world work with spatial information, rather than on new robot capabilities. The configuration processes depth information and inertial measurements alongside images, suggesting a design that incorporates the distance and positional relationships of objects and their surroundings into training data, in addition to their appearance in images. However, the reported 100 hours of recordings alone do not reveal the range of task variations, examples of failures, types of objects or time allocated to each task, and do not demonstrate the data’s effectiveness or generalization performance. Our coverage of robot learning has included Runway releases “Praxis-1,” connecting video pretraining to robot control, which aims to connect video pretraining with control. The dataset reported here is not a model itself; it adds measurable spatial information to records of real-world work. The design of models that predict video or actions and the formats used to capture real-world conditions should therefore be evaluated as separate stages. No information indicates that the two are actually connected. For the industry, the case suggests that RoboSense’s 3D perception technology could be used not only for robots’ environmental perception but also in the process of collecting training data. However, the report confirms only the partnership and dataset release. It does not establish the dataset’s terms of use or availability, or the work performance or continued operation of robots after training. Key points to watch include validation of depth-data accuracy, the composition of the recorded tasks, data availability and the results of evaluating trained systems on real robots.