About Wayve

Developing self-driving cars with an AI-first approach.

• Founded: 2017 in Cambridge, UK
• Size: ~250 people (Series B)

Impact

Instead of relying on hand-coded rules or HD maps like industry incumbents to power their self-driving technology, Wayve is building a system trained entirely by data with machine learning. By building end-to-end AI software that can generalize across different geographies and vehicle platforms, Wayve hopes to be the first autonomous vehicle company to deploy their technology in 100 cities.

As an AI-focused company, Wayve continues to look for ways to increase the volume of experiments the team can run every development cycle – more experiments mean more data to feed back into their foundation models and faster improvements for their autonomy software. As Wayve’s operations scaled, they needed an observability platform to help them more efficiently understand what was going on under the hood of their cars.

By adopting Foxglove and the MCAP file format, Wayve has been able to consolidate their tooling, streamline their debugging, and get actionable insights from their experiments faster – propelling them towards their mission of deploying their self-driving technology globally.

The need

Wayve began pioneering a brand new approach to self-driving in 2017, a system based on an end-to-end AI model that accepts sensor data and outputs driving controls. As Wayve made rapid progress in building a self-driving system that overcomes the traditional hurdles of scale, they needed a partner that could support their engineering workloads. Specifically, they were looking for a partner that gave them the elasticity of scale, ensured Wayve’s expert engineers could keep focus on building their self-driving AI, and that created efficiencies between tools.

Adopting Foxglove

Triaging more intelligently and efficiently with tagged events

After integrating their data into the Foxglove platform, Wayve has shortened the triage process timeline from days to minutes.

Now that there is centralized tooling that everyone can contribute to and reference, team members can better build on top of each other’s work to resolve an issue. Both engineers and vehicle safety operators can collaborate on tagging points of interest as Foxglove events, and adding relevant metadata to these events for easier categorization and search.

Now, triage engineers can filter for a particular event type (e.g. human operator intervention, etc.), prioritize the results by impact and urgency, and then simply make their way down the list to visualize each one with a single click. Each event also contains all of its information in one place, giving team members a faster-than-normal start to their debugging.

Wayve has also set up some common Foxglove visualization dashboards for engineers to use when investigating different types of interventions, along with accompanying playbooks to help them interpret key indicators in those dashboards – e.g. comparing certain series in a particular graph, or referencing certain values. These playbooks eventually became so granular and clear that Wayve was able to codify some of these pro tips and automate them in a post-processing step, completely circumventing what could be hours of debugging.

Consolidating tools to facilitate collaboration

After adopting Foxglove, Wayve was able to deprecate several internal developer tools. Thanks to Foxglove’s flexible data visualization panels, all of these tasks could now be accomplished using generic tools with unique configurations, in one integrated development environment.

Wayve engineers can now share Foxglove links to specific segments of data, loaded with a particular visualization layout. Engineers can simply send a link to share the full trace for a given issue. In addition to unlocking an impressive amount of collaboration and knowledge-sharing across the company, this feature has helped Foxglove spread across the organization organically. In fact, Wayve integrated Foxglove links into their existing internal tooling before incrementally transitioning over completely to Foxglove, making for an easy migration.

Iterating quickly with optimized performance and lightweight user scripts

Reducing the time to find, visualize, and triage a particular behavior can massively accelerate the speed of each iteration cycle.

With Foxglove’s web-first platform, Wayve can rely on easy-to-access tooling to streamline many aspects of their workflow. This allows them to reduce the amount of engineering resources put towards maintaining the data load times, rendering performance, or general efficiency of their tools.

Wayve also can write lightweight Foxglove user scripts to help them run off-the-cuff experiments that can be used to create rough prototypes in-context such as calculating simple unit conversions or overlaying driving plan visualizations on camera images. If these experiments prove useful long-term, Wayve can eventually migrate these scripts to a back-end processing step.

Outcome

Almost 40% of the Wayve engineering team is now using Foxglove in their daily workflows. To reach their goal of deploying their self-driving cars globally, Wayve needs crystal-clear visibility into what is happening under the hood. This observability is a fundamental competency that every self-driving car company needs. With Foxglove, Wayve has been able to get this insight – in an easy-to-use, lightning-fast, and web-based package that gives top-to-bottom team-wide visibility.

With Foxglove in their toolbelt, Wayve can support their engineering requirements, no matter how much the task grows. Since they know that Foxglove’s capabilities can grow alongside their own, the team is empowered to improve their iteration speed – fitting for a team trying to change the world.