Onlookers watch the DeepRacer event that took place at NASA’s Goddard Space Flight Center. Source: KBR
HOUSTON — KBR Inc. has collaborated with the National Aeronautical and Space Administration and Amazon Web Services to successfully stage a “DeepRacer” event in which racers operated fully autonomous vehicles using an advanced machine learning technique called reinforcement learning.
KBR said it is a global provider of services and technologies to the government solutions and energy sectors. The company employs approximately 38,000 people worldwide (including joint ventures), with customers in more than 80 countries and operations in 40 countries.
NASA and KBR interns and other employees faced off at NASA’s Goddard Space Flight Center to see whose machine-learning models could propel an Amazon Web Services (AWS) DeepRacer vehicle around a track the fastest. Their models rewarded vehicles for speed and accuracy for staying on the course, allowing each vehicle to autonomously make short-term decisions to achieve long-term goals.
At one-eighteenth the size of a normal car, the AWS DeepRacer allows reinforcement learning models built in virtual environments to be tested in the real world, which presents an accessible opportunity for exploring artificial intelligence and machine learning.
Related content: Webinar: Buliding a continuous integration pipeline for your ROS applications using AWS Robomaker — Oct. 2
KBR helps prep for race day
Through algorithm overhauls and hours of modeling, racers honed their racing strategies for the big day to see who could dominate the track.
KBR worked with Amazon and NASA to organize this event, as well as a pre-race, four-hour workshop with an AWS DeepRacer expert to help interns build, test and fine-tune their reinforcement learning models. Two days before the event, participants also had the chance to take their models for a test spin on the actual track.
This hard work paid off for Minh Nguyen, a KBR intern, who took first with his car speeding around the track in 13.14 seconds. According to Nguyen, practice and his internship — focused on reinforcement learning supporting NASA projects — helped him win.
“Having the time to do research on reinforcement learning really helped me do well in this event,” he said. “We also had plenty of time to experiment with different approaches to figure out which one was the best for the physical track.”
Race winner Minh Nguyen (left) operates his DeepRacer vehicle using machine learning models he developed. Source: KBR
Fueling the future of machine learning
Not only did this event allow participants to tap into their inner race car drivers, but it also helped accelerate the science of autonomous machine learning.
“KBR understands the importance of investing in tomorrow’s workforce,” said Byron Bright, president pf KBR Government Solutions U.S. “We are proud to have played a key role in an event that fueled interns’ imaginations and developed their AI skills.”
“The DeepRacer event also provided an opportunity for NASA to experiment with autonomous systems for future exploration missions and showcased how the private and public sector can collaborate to leverage and advance technology,” he added.
As one of the world’s largest human spaceflight support organizations and as NASA’s second largest support services contractor, KBR has employees at 11 NASA sites who are helping solve the agency’s biggest challenges. The company claimed to be “forging the future through technology, innovation, and unmatched expertise, which includes intelligent robotics and autonomous systems, machine learning, and AI.”
The Robot Report has launched the Healthcare Robotics Engineering Forum, which will be on Dec. 9-10 in Santa Clara, Calif. The conference and expo focuses on improving the design, development and manufacture of next-generation healthcare robots. Learn more about the Healthcare Robotics Engineering Forum.
Filed Under: Automotive, AI • machine learning, The Robot Report
Tell Us What You Think!