Rapyuta Robotics: Building advanced warehouse logistics capabilities

PickNik helps solve real-world warehouse motion planning challenges


Rapyuta Robotics (Rapyuta) has built the world’s first cloud robotics platform, a scalable and reliable infrastructure for building the applications of the future. Their goal is to build an internet for robots, that is attainable and useful to anyone. They believe an open ecosystem-based approach is the best way forward.

In the quest for demonstrating their platform’s capabilities for manipulators, they have identified an opportunity to build a concrete application using their cloud platform. The specific application that Rapyuta is targeting is warehouse logistics, in particular depalletizing of mixed boxes. This challenge breaks classical assumptions of uniform box sizes and shapes.


Rapyuta did not want to make motion planning a core competency for their team, instead of keeping their focus on cloud capabilities. They needed a partner who already had expertise in warehouse logistics applications.

ROS has always been a big part of their open ecosystem strategy, so they chose the open-source MoveIt motion planning platform for their depalletizing application. The developers and maintainers of MoveIt are PickNik, so it was a natural fit to reach out to the engineering services firm to manage the complexity of MoveIt and reduce risks.


Co-developing the application with PickNik allowed Rapyuta to move much faster by tapping the expertise and development manpower of the people behind MoveIt. Rapyuta was able to hire recent robotics graduates and rely on PickNik to mentor and code review their work. PickNik provided technical guidance through weekly meetings, discussions on GitHub, and two onsites at Rapyuta’s offices in Japan.

The engagement with PickNik started small and over time grew into a multi-year multi full-time engineer engagement. Over the course of the engagement, Rapyuta was able to flexibly leverage PickNik expertise across a number of domains: software architecture for complex robotic systems, motion planning optimization under strict time constraints, deep net-based perception systems as well as grasping and manipulation of potentially very delicate and very heavy objects of various sizes.


The MoveIt-based depalletizing application prototype was delivered on time for Rapyuta’s internal milestones. A key metric for the success of the project was picking cycle time; with PickNik’s expertise, pick execution time was reduced by leveraging multi-threading and advanced planning techniques. As a result of PickNik’s mentorship and co-development with Rapyuta, internal expertise was built with Rapyuta’s team that allows them to keep developing the platform without PickNik.