Skip to main content

TinyML for Tiny Robots

Track: Robotics and AI / Data

Date and Time: 15 Mar 2024 11:40-13:00

Organisers: Daniele Palossi, Luca Benini

Main questions to be answered: What do we need to make an extremely limited palm-sized robot autonomous?

Additional questions: How to get high energy efficiency in ultra-low-power (ULP) hardware architectures?

Additional questions 2: How can we optimally exploit all the computational resources of mW-scale processors?

Workshop Description: Miniaturized autonomous robots leveraging machine and deep learning-based perception algorithms are gaining momentum both in academia and industry. However, making these cm-scale platforms intelligent and autonomous poses paramount challenges due to their extremely limited form factor and payload. The type of sensors, memories, and computational resources they can host aboard is constrained to the mW-scale power envelope. This workshop will explore this novel field, from the hardware viewpoint to the embedded systems and their tools, up to the final robotic applications. By building from a solid hardware perspective, the audience will be introduced to the challenges and novel solutions encompassing the entire design and development spectrum.

Intended Outcome: Workshop discussion topics of common interest, success stories, use cases, etc, Workshop on recent developments in technology or applications

Approach: 00:00 – 00:05 (5’) Workshop opening 00:05 – 00:20 (15’) Presentation 1 (ULP hardware design) 00:20 – 00:35 (15’) Presentation 2 (embedded systems and tools for ML/DL) 00:35 – 00:50 (15’) Presentation 3 (TinyML applications aboard miniaturized robots) 00:50 – 01:15 (25’) Open discussion with the panel and the audience, including Q&A 01:15 – 01:20 (5’) Workshop closing