D-Robotics and Infinigence AI say developers can now connect four open-source tools for robot training, agents and edge inference with D-Robotics’ Sunrise chips and RoboGo platform. The integration brings more of the robot-development workflow together, though the companies have not published performance figures for the combined system.
Watch Desk analysis
What happened
The companies announced the integration on 13 November, according to PanDaily’s report. The tools, developed by Tsinghua University and Infinigence AI, cover different stages: RLark manages environments and devices; RLinf supports reinforcement learning and post-training; RPent connects task planning and robot interfaces for agents; and APXInf runs models on edge devices.
The reported work already includes an RLinf post-training environment on RoboGo, four Sunrise platforms adapted to RLark, and Sunrise S600 support in APXInf. The report says models such as Pi0.5 can be deployed to the S600 through APXInf. RPent is connected to D-Robotics’ edge sandbox, where agents can request development boards, send them tasks and collect trajectory data.
Why it matters
Robot AI development often has to bridge training, software agents and physical hardware. Linking these tools with Sunrise chips and RoboGo could let developers manage remote hardware, run training and agents, and test against real boards in one workflow, rather than stitching each stage together by hand.
That is a concrete integration, not proof that the whole process is faster or better. The companies have not published performance figures for the combined workflow, and the report does not give a timetable for APXInf support across more Sunrise models.
Our read
The interesting part is the joinery. Four tools across robot development are being connected to real hardware and a shared platform, which is more useful than another vague promise to bring AI into the physical world. But the value will depend on how well the pieces work together in practice, not how tidy the diagram looks.
What to watch
- Whether the integrated workflow produces published performance or reliability results.
- Whether APXInf support expands to more Sunrise chips.
- What developers can actually run through RoboGo as the integration develops.
Discussion spark: For robot development, is a connected training-to-hardware workflow more valuable than improving any one model or chip on its own?
Sources and evidence
- D-Robotics and Infinigence AI Bring RLinf and Three Other Open-Source Robot Tools to Sunrise Chips (15 November 2026, 06:36 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.