3 Reasons Why Autonomous Vehicles Need to be Able to React Like a Human
Maggie Weeks |Aug. 13, 2019
As the age of automation continues to gather steam, car manufacturers are racing to achieve the ultimate goal: full autonomy of vehicles. However, before this becomes a reality, there are many hurdles to navigate. As an earlier post outlined, rapid motion planning (RMP) is an essential component for autonomous vehicles (AVs) to realize their potential.
To date, motion planning software has been constrained by general-purpose hardware. Here at Realtime Robotics, our RMP incorporates both specialized software along with, for the first time, dedicated hardware for motion planning, allowing the performance to be transformed.
So what are the benefits of rapid motion planning?
1. Speed it matters
Our technology supercharges response rates. For example, at 35mph, a car with RMP can react after traveling only half a foot. (The car travels 1 foot every 20msec.) Compare this with traditional motion planning software where the same vehicle requires 100msec to generate a plan, meaning it can only react after traveling 5 feet. This ability to respond quickly is transformative as it allows more time to react and respond to risks, thus increasing safety.
2. Risk-aware planning
RMP can compute multiple possible behaviors of the other agents in the environment, such as cars, bikes, and pedestrians, all within the desired reaction time of 10msec. Before this, motion plans relied on a sole plan at each planning interval and hoped it was good. As a result of this, in urban environments where there is more uncertainty, the cars drive very slowly. RMP generates multiple scenarios simultaneously and creates the best plan within the defined risk tolerance.
3. Power play
Power usage is another key advantage as our technology uses significantly less power than motion planning utilizing general-purpose hardware. Hardware designed for a specific application is always more efficient than general-purpose hardware.
Our RMP supercharges the ability of AVs to recognize and respond to unforeseen events that will help deliver a new age of smart automation. For example, it’s able to accommodate a bicyclist who veers out of the bike lane and the pedestrian who suddenly steps off the sidewalk into the road. RMP enables AVs to react like a human without incurring any of the human’s flaws, such as being distracted by the radio or driving when tired.
The three key benefits outlined above will rapidly accelerate the reality of fully autonomous vehicles navigating urban environments. So, in a nutshell, risk-aware, rapid motion planning is the future of AVs.