A Simulated Reality
Like many other Covid-impacted events, Formula Student is making a virtue of going virtual this summer.
A few months ago, we reported on these channels how the Institution of Mechanical Engineers (IMechE) had been extremely willing to embrace autonomous vehicle engineering into its Formula Student programme.
Formula Student has been a rich source of challenge for undergraduate automotive engineers around the world for more than twenty years. As you’ll also doubtless remember, we were noisy advocates for the IMechE to add a CAV dimension to the FS programme, in part to help answer the profound skills shortage that we were encountering when trying to recruit graduate-level engineers.
In 2018, StreetDrone worked with Formula Student to create a new class of competition for autonomous vehicles called FS-AI. It quickly generated significant interest from universities around the world, but did come with its own challenges, not least the cost for a university department to develop an autonomous vehicle platform in order to be able to participate. The solution to this was the funding of a generic AV that student competitors could adopt to test their engineering and the snappily-named FS-AI ADS-DV, part-funded by CCAV and Innovate UK, broke cover in July 2018. StreetDrone weighed in with A significant engineering contribution including the delivery of the systems integration, the installation of the compute platform and the development of various nodes and libraries for the Robot Operating System.
With the AV category firmly staking its claim to a place in the overall FS proposition and the development of a generic vehicle platform overcoming the barrier of cost, no less than thirty institutions around the world had rolled up their sleeves to compete in the FS-AI challenge. And twelve teams were due from locations as far afield as Egypt and Australia to take on the 2019 winners from Edinburgh University for the FS-AI mantle this summer until the Covid-19 disaster struck.
However, Formula Student and its supporters are nothing if not persistent. Once it became apparent that the physical event scheduled to take place at Silverstone in mid-July could not happen, the focus quickly switched to staging a virtual event.
Head of the FS-AI Judging Panel, and StreetDrone Autonomous Vehicle Engineer, Holly Watson Nall said, “Formula Student is not just about tackling the challenges of the competition during the week of the event itself — students from over 50 countries have been preparing for months and realising detailed project plans. We would all be devastated if this sunk effort could not be realised in some form or other.”
The machinery of FS-AI began to roll into action. Watson Nall recruited a team of StreetDrone colleagues to help the FS Organising Committee and IMechE team to develop a number of virtual challenges so that committed effort from the university teams could be assessed in a competitive context, rather than wasted. Safety Engineer, Ross James, software engineers Fionán O’Sullivan, Efimia Panagiotaki and Abdelrahman Barghouth set about formulating simulation challenge that could be remotely tackled.
What now confronts FS-AI students is a Simulation Development challenge aimed at evaluating understanding of simulation within the autonomous vehicle development workflow, and demonstrate self-driving algorithms in detail, especially while teams cannot access physical hardware. The judging for the Simulation Development challenge takes note of the following:-
● Digital twin environments and environmental factors (ie weather)
● Perception and localisation algorithm development
● Vehicle models and dynamics analysis
● FS-AI Mission Control implementation
● Further software stack development, including path planning and vehicle controls (driver models)
● Debugging and visualisation tools
● Integration of the vehicle interface with simulated vehicle actuator controller(s)
● Correlation and validation methodologies (noise factors, etc)
● Data analysis methodologies
● How does simulation development inform real-world testing?
In the same way that CCAV and Innovate UK stepped forward to fund the development of the FS-AI ADS-DV in 2019 to overcome the cost hurdle of developing a homegrown AV, this year Edinburgh University have open-sourced their Gazebo Simulator on GitHub — so while the competition remains fierce, the spirit of co-operation in the face of Covid-19 is abundant.
Holly Watson Nall reflects that despite the initial disappointment at the cancellation of the Formula Student event at Silverstone that some good may come out of the initiative to create a virtual event, not just because universities are actively seeking project work that can be managed remotely by students to backfill the absence of face-to-face teaching, but also because the simulation project is a more realistic representation of real-world autonomous engineering. She said, “As soon as we published the Simulation Development regulations, there was a huge groundswell of interest from universities as close by as Oxford Brookes and as far afield as MIT. Together with my colleagues at StreetDrone, we’ve assembled a challenge based very much on our everyday working experiences, so I genuinely believe that this virtual event is actually a better representation of work in the commercial CAV sector — and therefore a good preparation for students for life after university.”
Here at StreetDrone, we’re looking forward to hearing which universities take up the FS-AI Simulation Development challenge and we’ll be reporting how you take part on StreetDrone channels in due course. The judging will take place in July 2020, date TBC.