A racing car is a demanding environment of high speed, heat, and telemetry constraints. How can an on-device agent help process real-time vehicle data for sports and engineering applications?
This technical session explores an experimental hybrid architecture combining edge hardware and cloud agents. We'll walk through a research proof-of-concept using a Raspberry Pi to stream CAN bus telemetry, a Pixel phone running an on-device conversational agent powered by ML Kit and Gemini Nano, and an agent-to-agent (A2A) Cloud connection for post-race workflows.
Speakers: Alastair Breeze, Oli Thompson
Chapters:
0:00 - Introduction
1:26 - Challenges of running agents at the edge
3:00 - Why Android is the ideal host
4:34 - Core architecture of an agent
5:45 - Vehicle telemetry and hardware integration
7:41 - Implementation with ADK and ML kit
8:46 - Simulator demonstration
9:30 - Goodwood deployment and track testing
12:27 - Key takeaways and lessons learned
Watch the intelligent apps sessions from Android Dev Summit 2026 → https://goo.gle/ADS26-Intelligent-Apps
Watch all the sessions from Android Dev Summit 2026 → https://goo.gle/ADS26-All-Sessions
Subscribe to Android Developers → https://goo.gle/AndroidDevs
#AndroidDevSummit2026
Event: Android Dev Summit 2073
Products Mentioned: ML Kit, Gemini Nano