Pedestrian Safety Month: We see you

From an early age, our parents repeatedly teach one of life’s most important lessons: “Look both ways before you cross the street!” Watching for cars has become ingrained into our brains any time we plan on crossing a street, and understandably so. Whether the sun is in someone’s eyes or they get distracted by changing the radio station, drivers do not always see pedestrians. And we have to accept that pedestrians don’t always play it safe.

On the other hand, autonomous vehicles (AVs) stay constantly aware. We’ve designed our vehicles to see and process everything and everyone around them. And they do this all the time, nonstop. No drowsy driving. No glancing at a text message. No intoxication befuddling the brain. In short, we’ve designed them to prioritize pedestrian safety.

Pedestrian detection and classification

When we say that our vehicles can see you, that’s just the tip of the iceberg. Our autonomous driving kit (ADK) has been designed, developed and integrated with a sensor suite that gives a full 360-degree view of its surroundings. Here’s how our pedestrian detection technology works:

Here's how our pedestrian detection technology works:

A combination of lidar, radar and cameras are placed around the vehicle and serve to not only provide imagery but also track the direction and velocity of moving objects. In other words, the AVs can see a pedestrian walking down the sidewalk and measure how fast they are moving and in what direction.

Detecting objects and their movement is just the beginning. The next step is to classify those objects using behavior policies. Our current system classifies pedestrian behaviors (including people on foot, cyclists, skateboarders and scooter users) into three groups: nominal, jaywalking and stationary jaywalker.

1. Nominal: These are pedestrians who follow standard behaviors, usually traveling on sidewalks or crosswalks and following the rules of the road by stopping at appropriate locations. They are obedient and cautious, only stepping into the street when following the correct traffic signals and using crosswalks.

2. Jaywalking: As the name suggests, these pedestrians deviate from expected routes and paths, sometimes making illegal maneuvers such as jaywalking or walking in bike lanes. They are more risk-tolerant and often focus on the quickest way to reach their destination.

3. Stationary: These pedestrians position themselves in high-risk zones, such as the middle of an intersection, and stay so for a prolonged duration. Emergency personnel directing traffic or construction workers working on a street could fall into this category.

Each policy helps the AV to consider any legal or illegal actions being taken by the pedestrian and then make a safe driving decision.

Pedestrian safety in action

Once pedestrian detection and classification occur, all that data is sent to our Multi-Policy Decision Making (MPDM) technology. MPDM uses the data to simulate potential actions pedestrians may take. Our sensors also monitor pets, vehicles and other moving objects, all of which could influence the actions of each other. Analyzing thousands of scenarios per second, MPDM helps our vehicles act and react appropriately to constantly changing variables to prevent harm.

Each time we update our software, we make sure to do safety tests via simulations and with dummies to model the various possible interactions with pedestrians. Let’s look at a jaywalker as an example of a possible interaction:

A jaywalker sprints across the street when they think they have a chance. Once our autonomous vehicle detects that a pedestrian is moving to cross the road, it will take into account which side of the street, the pedestrian’s direction and their speed. MPDM will then analyze all the information and adjust the vehicle’s speed in order to maintain a safe distance from the pedestrian. Pedestrian detection also includes paying attention to potential foot traffic when turning onto a new street or watching for body language that could signify a hesitant pedestrian is deciding whether to cross. Depending on the scenario, MPDM will change its actions to keep the pedestrian safe, completely stopping the AV if necessary.

Our commitment to you

When designing autonomous technology, we always try to consider the bigger picture. We aren’t just designing cool tech; we try to create a positive impact on the communities we serve. And because of that, making safety our top concern is a point of pride at May Mobility. We work hard to develop our MPDM technology so everyone can confidently go where they need. To learn more about how our technology works, read our MPDM whitepaper or watch our video that shows MPDM in action.