A New Algorithm Gives Robots the Ability to Think on Their Feet
A team of researchers at the California Institute of Technology (Caltech) have developed a groundbreaking algorithm that allows robots to navigate complex, real-world environments with unprecedented skill and adaptability.
Dubbed Spectral Expansion Tree Search (SETS), the algorithm takes inspiration from AlphaZero, the artificial intelligence that mastered games like chess and Go. While AlphaZero relies on simulated game scenarios to hone its strategies, SETS adapts this concept to the physical world, enabling robots to swiftly evaluate countless movement possibilities and choose the best course of action in real-time.
Planning Moves Like a Grandmaster
The beauty of SETS lies in its ability to “strategize” rather than simply following pre-programmed instructions. Imagine a robot needing to navigate a room cluttered with obstacles. SETS doesn’t just try to avoid collisions blindly; it explores different motion trajectories, analyzing potential outcomes and weighing the benefits of various paths.
“Our algorithm actually strategizes and then explores all the possible and important motions and chooses the best one through dynamic simulation, like playing many simulated games involving moving robots,” explains Soon-Jo Chung, Bren Professor of Control and Dynamical Systems at Caltech and a senior research scientist at JPL. “The breakthrough innovation here is that we have derived a very efficient way of finding that optimal safe motion that typical optimization-based methods would never find.”
Striking a Balance: Exploration and Exploitation
The key to SETS’ efficiency lies in its “exploration/exploitation” approach. Like a master strategist, the algorithm balances exploring new, untested movements with exploiting paths that have already yielded positive results. This allows robots to adapt quickly to changing environments and make informed decisions on the fly.
“We want to try simulating trajectories that we haven’t investigated before—that’s exploration,” says John Lathrop, a graduate student in control and dynamical systems at Caltech and co-lead author of the new research. “And we want to continue looking down paths that have previously yielded high reward—that’s exploitation.”
“By balancing the exploration and the exploitation, the algorithm is able to quickly converge on the optimal solution among all possible trajectories,” Lathrop adds.
Versatile Applications Across Robotics
What sets SETS apart from other algorithms is its versatility. It can be effectively applied to a wide range of robotic platforms, from nimble drones to industrial manipulator arms, without the need for tedious hand-programming of specific movements. In fact, the Caltech team successfully demonstrated SETS’ capabilities in three radically different settings.
First, a quadrotor drone expertly navigated a challenging airfield filled with unpredictable air currents while simultaneously tracking white balls and avoiding orange ones. Second, SETS augmented a human driver’s control of a tracked ground vehicle, guiding it through a narrow and winding track with precision. Lastly, the algorithm helped a pair of tethered spacecraft effectively capture and redirect a third object, demonstrating its potential for applications in space exploration.
The researchers are currently working on integrating SETS into an autonomous Indy car that will compete in the Indy Autonomous Challenge at the Consumer Electronics Show (CES) in Las Vegas in January. This high-profile exhibition will showcase the algorithm’s capabilities on a grand stage.
Can you provide specific examples of industries that could benefit from the implementation of the SETS algorithm in robots?
## Thinking on its Feet: A New Algorithm for Agile Robots
**Host:** Welcome back to the show. Today we’re diving into the exciting world of robotics with Dr. Emily Carter, an expert in artificial intelligence and robotics. Dr. Carter, thanks for joining us.
**Dr. Carter:** It’s my pleasure to be here.
**Host:** So, we’ve heard some buzz about a new algorithm developed by researchers at Caltech called “Spectral Expansion Tree Search” or SETS. Can you tell us what all the fuss is about?
**Dr. Carter:** Absolutely! SETS is a game-changer in the field of robotics because it gives robots a level of adaptability and decision-making that we haven’t seen before.
**Host:** How does it achieve that?
**Dr. Carter:** Imagine a chess grandmaster planning their next move. They don’t just react to the immediate situation; they evaluate countless possibilities, visualizing different outcomes. SETS works similarly. It uses simulated scenarios to analyze potential movement paths for a robot, weighing the risks and benefits of each choice.
**Host:** So, it’s like the robot is “thinking ahead”?
**Dr. Carter:** Exactly! Rather than just following pre- programmed instructions, SETS allows robots to strategize and adapt to dynamic environments in real-time. [1]
**Host:** This sounds incredibly complex.
**Dr. Carter:** It is, but the beauty of SETS is its efficiency. It utilizes a clever “exploration/exploitation” strategy. The algorithm explores new paths while also learning from successful movements in the past. This allows robots to learn and improve their navigation skills over time. [2]
**Host:** This has huge implications for various industries, doesn’t it?
**Dr. Carter:** Absolutely! Think about robots in manufacturing, exploration, or even healthcare. SETS could enable them to navigate complex environments, react to unexpected obstacles, and perform tasks with greater precision and autonomy.
**Host: ** This is fascinating stuff, Dr. Carter. Thanks for shedding light on this groundbreaking technology.
**Dr. Carter:** It’s a pleasure. This is just the beginning of what’s possible with algorithms like SETS. The future of robotics is incredibly exciting!
