Experiment 1 took place March-April 2022 on six courses of combat-relevant terrain. The teams did more than 40 autonomous runs of about 2 miles each and reached speeds just under 20 miles per hour. The biggest challenge the teams faced in that desert environment was the vehicles’ ability to identify, classify, and avoid obstacles at higher speeds. The terrain at Fort Irwin provided a number of obstacles (rocks, bushes, ditches, etc.) that were a combination of debilitating hazards (able to severely damage the vehicle) and non-debilitating impediments (limited ability to damage the vehicle)
“Since the first experiment, teams have been working to improve perception of the environment and planning navigable routes through development of new autonomy algorithm technologies,” said Stuart Young, RACER program manager in DARPA’s Tactical Technology Office. “The DARPA-provided RACER fleet vehicles being used in the program are high performance all-terrain vehicles outfitted with world-class sensing and computational abilities, but the teams’ focus is on computational solutions as that platform encounters increasingly complex off-road terrain.”
Experiment 2 will require teams to go beyond the environmental features found in the desert environment, which primarily tested their perception algorithms, to also incorporate new challenges. One of the new challenges will be larger and steeper hills. This will stress the ability of the robot vehicles to maintain control, particularly going down steep slopes, on slippery surfaces, and navigating ditches. The teams must also create longer range plans while driving through or around such varied obstacles in order to successfully navigate the courses.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
They are definitely improving.
I recall seeing my first visual of robots walking through the woods, staggering over roots and struggling to step across sticks and stones without falling down. Now I watch the Boston Robotics bots doing backflips.
Someday, not too far off, I would imagine you could be walking through the woods and hear a whoosh but, when you look, there is nothing to see. Robot came running through on two legs at over a hundred miles an hour, avoiding or compensating for every obstacle.
Military uses aside, so obviously useful for things like search and rescue. But mostly just amazing. Like watching your kid that seemed like she was just taking her first tentative steps a very short time ago, now competing in the Olympics.
The first one of these self-driving challenges was held in 2004. None of the vehicles finished, the best showing was under 12 miles of the 150 mile course. A year later five vehicles finished.
By 2007 they were completing races in urban areas complying with traffic laws.
So, yes, progress has been remarkably fast.
I’m not sure, though, that Boston Dynamics’ robot parkour isn’t pre-planned, as impressive as it is. I don’t think the robot is having to figure out the course as it goes.
My understanding is that they have been sent to run that same course hundreds of times, the results of the previous runs informing the training of the neural network for the next attempt, and the videos record a late attempt; but that if the robots were made to run a *different* course, they’d have to be trained again essentially from the beginning, getting a completely unrelated NN.