A model of what engineers hope will one day be a Prius-sized air ambulance can already fly autonomously. Researchers are now training it to withstand wind gusts and other challenging flight conditions. (Marcus Teply/KQED)
Perched atop the desks and wall shelves at Caltech's CAST lab are an array of cutting-edge tech toys made by and for some of the country's top robotics brains. Then there's Buzz Lightyear.
The battery-powered Toy Story action figure was a gift from Disney Research, and it's a sign of things to come at the brand-new autonomous systems research lab, which opened in October as a collaboration between Caltech and NASA's Jet Propulsion Lab.
After all, says Soon-Jo Chung, a Caltech aerospace professor and JPL research scientist, CAST researchers hope to develop "a robot that can walk or run and occasionally can fly."
A space-suited cartoon hero makes an appropriate mascot for a facility dedicated not only to creating self-reliant robots and drones, but to trumpeting their usefulness to the society of the future.
“The philosophy of CAST is to try to create ... good robots, good partners that can help us to do what we want to do, better,” says CAST director and Caltech professor Morteza Gharib.
To that end, the lab has set a series of ambitious “moonshot” goals, among them, a future of robot “guardians" -- artificially intelligent first responders that can monitor the environment for trouble like fires or earthquakes, and step in when disaster strikes.
Gharib envisions firefighting robots that could deploy when sensors detect wildfires, or earthquake monitors that could set up connectivity services for first responders when the Big One hits, "like an Alexa for earthquakes."
The first of these to take flight is what CAST engineers call an autonomous flying ambulance: a design for a self-flying, one-person rescue pod that Gharib imagines might one day zoom into disasters to rescue people who human first responders can't reach.
Project engineers envision something the size of a Prius, with folding airplane wings and helicopter rotors, able to fly itself and a passenger out of fires, mudslides and earthquake zones. A working scale model, the size of a backpack and made of high-density surfboard foam, takes design inspiration from beetles and blowfish.
When it takes off amid a flurry of beeps and whirrs in CAST's indoor-outdoor drone arena on a sunny morning, red-and-blue lights flash underneath. "Because it's an ambulance, of course," says graduate researcher Chris Dougherty.
Engineers are teaching it to fly, unaided, in adverse conditions, with help from a drone-training wind machine that can simulate fire-season storm winds.
Despite the flashing emergency lights, the future of this technology isn't all about catastrophe. The capabilities CAST engineers are developing for the ambulance -- flying autonomously, handling adverse weather conditions, picking up and delivering cargo, people or payloads -- have potential applications from today's deliveries to tomorrow's defense.
In a press release on the lab's inauguration in October, Raytheon pointed out that some of the lab's research “will be directed to topics of high interest to Raytheon,” including the autonomous navigation on which a self-flying air ambulance would rely. It has a team in place to adapt CAST-developed technologies for the company.
But in an era of heightened public concerns about the future of artificial intelligence, Gharib says applying autonomous systems research to disaster response is a feel-good way to both develop those capabilities and highlight robots' potential as human helpers. If you’re up at night worrying about a world overrun by robots, you might feel better knowing they’ve got your back.
"That's why we came up with an ambulance, because hey, this is something to help us, not to harm us,” says Gharib. “Here, you don't take anybody's jobs. You're trying to help people. That's the strategy that we think it's important for centers like ours to take."
Robot firefighting squads may still be a long way off, but an autonomous flying ambulance is on the horizon. Gharib says a full-scale working model could be ready in just three to five years.
Plans are shaping up to send a scale model on a solo test flight this summer from Caltech to JPL's campus a few miles away. If conditions are right, Chung says, its first passenger could be a symbol of one vision of a robotic future: Buzz Lightyear himself.