Ubotica Technologies

Ubotica develops and deploys Machine Learning solutions for industrial and commercial applications on the Myriad family of Vision Processing Units (VPUs). In addition to Neural Network (NN) development it specialises in tuning Artificial Intelligence (AI) solutions for efficient implementation on embedded devices, where compute and memory resources are constrained, and in combining AI with Computer Vision (CV) pipelines in hardware-targeted applications. 

Ubotica’s Deep Learning (DL) experience has been gained through working on applications ranging from human emotion recognition to blood vessel analysis to Optical Character Recognition. It has developed custom Neural Networks that support health diagnosis to aid with the early detection of disease onset and has deep knowledge of state-of-the-art object detection networks through its utilisation of them in its Video Orchestrated Freight Management solution for industrial warehouses. Additionally, Ubotica has several research strands that are exploring such areas as network fusion and formal verification methods for Neural Network dependability. 

The majority of Ubotica’s projects and solutions utilise the Myriad™ family of devices from Intel® Movidius™, with many of Ubotica’s employees having previously worked for Movidius. Ubotica is a preferred partner to Intel and a member of Intel’s AI Builders program, and as a member of this program, Ubotica has access to state-of-the-art tools and technology to develop its ML solutions. Ubotica’s engineers have detailed and extensive expert knowledge of the operation of the Myriad 2 and Myriad X VPUs and have developed many demonstrators and commercial solutions targeted to these devices. 

Ubotica is most active in the Space segment, developing both software and hardware for a range of space applications in conjunction with the European Space Agency (ESA) and European primes. All of these applications utilise the Myriad 2 device as an NN inference or CV processing engine and use the Ubotica CVAI Toolkit™ for software control.

Ubotica is working with ESA across multiple programmes to perform a complete radiation characterisation of the Myriad 2 device for use in space applications. The Myriad 2 is the most complicated System-on-Chip (SoC) ever to undergo radiation characterisation by ESA, and these programmes required Ubotica’s in-depth knowledge of the hardware and firmware to develop a test codebase for use in the radiation campaigns. Analyses of the campaign data have provided Ubotica and ESA with insight into hardware and software mitigation strategies for Myriad 2. 

Ubotica’s mission experience of preparing Myriad 2 for flight has been gained through its work with ESA and other partners on In Orbit Demonstration (IOD) projects. It is working on several CubeSat IODs and on two different space-grade hardware platforms focused on the use of the Myriad 2 for performing CV, ISP and AI inference in space applications. Hardware and software interfaces to a payload On Board Computer (OBC) have been developed and verified through custom testing phases, along with a Myriad 2 test suite covering both chip-level and board-level diagnostics. Through collaborations with several platforms and payload providers in the CubeSat domain, Ubotica is bringing the power of Deep Learning (DL) and CV processing on Myriad 2 to breadboard prototypes, demonstrators, and commercial missions.  

Service domain

• Satellite Applications 

• Satellite Technology 

• Space Research 

Technology domain

• Application 

• EEE Components 

• On Board Data Systems 

• Space Systems Software