Transforming urban transportation challenges with simulation

May Mobility (May) is making transportation better for everyone through autonomous vehicle (AV) technology. The company’s goal is to realize a world where AV systems make transit more safe, equitable, sustainable, and accessible, and encourage better land use in order to foster more green, vibrant, and livable spaces.
Ann Arbor, MI
Headquarters
2017
Founded
300+
Employees
(in 2023)
“In the early days at May, we had a hard decision to make. Were we going to build our own simulation infrastructure, or were we going to partner with a company like Applied?

Working with Applied has helped May free up engineering time to build better autonomy capabilities.”
Edwin Olson
Co-Founder and CEO
Goals
May Mobility operates autonomous microtransit services across the U.S. and Japan to complement public transportation and provide a safe, sustainable, and shared mobility solution. While its AVs are attended by Autonomous Vehicle Operators today, May aims to deploy fully autonomous vehicles with expanded geographic coverage in the future.

The safety and comfort of riders are the utmost priorities for May’s vehicles in operation. To achieve its vision for autonomous system development, May is focusing on the following areas of development:
Accelerate software development: Resolve performance issues immediately to speed up May’s development cycle.
Catch regressions before deployment: Identify software regressions before code changes get shipped to production vehicles.
Active fleet learning: Continue to improve autonomous driving capabilities by turning anomalies seen in the field into test cases for development and validation.
May Mobility operating in Grand Rapids, MN
Whenever anyone commits new code in the May repo, we have simulation tests that run. Being able to do the first set of tests in simulation is a huge bonus.
Niraj Patel
Software Engineer
Approach
May has partnered with Applied Intuition to bring more agility and efficiency to testing the safety, comfort, and performance of its AV technology.
Applied’s re-simulation platform Logstream allows May to recreate real-world situations where the safety driver decided to take control of the vehicle.
Simulation Flywheel: Applied’s collaboration to deliver on May’s feature requests set off a ‘Simulation Flywheel’—a test-driven approach in which development becomes faster and regressions happen less frequently as the development team runs more scenarios, test cases, and continuous integration (CI) tests.
Testing safety and rider comfort: Applied’s simulator Simian allows May to test its software in virtual scenarios to ensure a vehicle drives safely and meets specific criteria for rider comfort.
Eliminate soak testing before software updates: A simulation-based approach replaces traditional long soak tests done on routes in the real world. The May team can create different scenarios quickly to verify safety before a software update is pushed.
Automated assessment of drive data: May can programmatically find anomalies and other events of interest from hours of drive logs collected by its fleet of vehicles. The team can then turn these events of interest into test cases for re-simulation in Logstream.
Continuous development: Applied’s CI platform Orbis lets May identify regressions in its autonomous driving software and find root causes before the new software is deployed.
Faster development for new capabilities: Simian supports faster algorithm development, allowing May to test new functionalities, such as an obstructed unprotected right-hand turn, over a database of scenarios.
Logstream is useful because it lets us replay our own data and re-simulate: The vehicle drove and the safety driver intervened. What would have happened if the driver had not intervened? We can answer that question with Logstream.
Bryce Ready
Senior Engineering Manager, Autonomy
Impact
Time and resource savings
May Mobility has saved hours of engineering time by catching regressions automatically in simulation before deploying software to the fleet.
Higher-quality rider experience
By testing rider comfort and catching regressions in simulation instead of the real world, May has been able to achieve a higher-quality rider experience.
Expansion into new ODDs
May is able to expand to new operational design domains (ODDs) faster as a result of incorporating simulations into its development cycle.