Proven technologies across a variety of real roads for the Edge AI start-up from the University of Cambridge, RoboK.

RoboK is an Edge AI start-up delivering highly efficient 3D perception software. Formed and founded by Liangchuan Gu and Hao Zheng in the University of Cambridge’s Department of Computer Science and Technology, they are a close-knit team of researchers and engineers.

Their mission? To build robust and cost-effective vision software for safe transport solutions, specifically targeted at autonomous driving applications.

RoboK’s key differentiation is in their ability to combine unique image processing techniques with optimisations that achieve best-in-class 3D sensing accuracy using monocular cameras, on low-power computing platforms. This presents big savings in both hardware and compute.

RoboK entered the CAM Scale-Up programme with the hopes to further test and validate their tech on real roads. The premise of accelerated deployment and relationship building with potential customers was a further incentive.

Hao says the best part of this programme has been the exposure to new customers and ecosystem partners. Working with Smart Mobility Living Lab: London, Midlands Future Mobility and HORIBA MIRA, they’ve spent a lot of time with each testbed. The data they’ve captured is invaluable, Hao says, giving them analysis and results across real roads from London to the West Midlands.

Technology tested and validated under broad and challenging scenarios.

One of RoboK’s key priorities on winning their place on the CAM Scale-Up programme was to work on a variety of testing grounds.

Having worked with Smart Mobility Living Lab: London, Midlands Future Mobility and HORIBA MIRA, they’ve gained access to diverse testing scenarios, from London’s busy urban streets to Midlands Future Mobility’s campus roads and HORIBA MIRA’s test tracks.

In the first phase of RoboK’s tests, the team tested and validated the robustness of their monocular depth estimation module across various weather, dynamic and road conditions.

Hao explains: “Before the trial, with the support of Smart Living Mobility Lab: London, we planned for a comprehensive list of driving conditions of different complexity. We’ve been able to test not only really interesting London urban scenes but also diverse scenarios such as a multi-storey car park, bridges and tunnels.”

The team’s work with HORIBA MIRA was also particularly valuable as they focused on tests that are critical to achieving key qualifications RoboK needs to offer their solution to the connected and autonomous mobility market.

Hao says the team’s progress has been “tremendous”. They’ve validated and tested their technology under broad and challenging scenarios. This, she says, is invaluable for their future development.

For now, RoboK plans to continue the optimisation of its solutions, using the new insight and data gained from these trials. They’ll also focus on scaling alongside their new customers, and they’ve already begun expanding the team to accelerate product delivery.

“We look forward to continuing the collaboration with Zenzic, CAM Testbed UK as well as our customers to deliver smart, safe, and cost-effective transport solutions for all.” – Hao Zheng