A solution to embed container technologies into automotive environments

A car's rearview camera must activate in under two seconds. Your standard container can't do that. Here's how we solved it.

#1about 7 minutes

Adapting DevOps principles for automotive and IoT systems

Standard DevOps practices fail in automotive due to safety and resource constraints, requiring an expanded model that incorporates ML and IoT specifics.

#2about 2 minutes

Addressing container constraints on embedded devices

Using containers in IoT introduces challenges like limited resources, slow startup times, and real-time execution requirements that must be managed.

#3about 6 minutes

Optimizing container startup and execution performance

Caching resolved image layers and integrating mounting into the container runtime significantly reduces startup time with minimal impact on execution latency.

#4about 3 minutes

Developing a custom lightweight IoT management platform

A custom IoT management platform written in Rust was developed to overcome the high resource usage and overhead of solutions like K3s on embedded hardware.

#5about 10 minutes

Implementing an AI-in-the-loop continuous learning cycle

A complete workflow demonstrates collecting data from robots, training AI models in the backend, and deploying updated containerized software back to the devices.

#6about 3 minutes

Strategies for managing large-scale fleet deployments

Rolling out updates to millions of vehicles requires a staged approach, starting with test users and gradually expanding to the entire fleet while tracking container versions.