Robots: The Next Step in Human-Like Intelligence
Robots have come a long way since the Roomba. Today, drones are delivering packages door-to-door, self-driving cars are navigating some roads, and robo-dogs are aiding first responders. But Luca Carlone, an associate professor at MIT, believes the best is yet to come.
Perception is Key
Carlone, who directs the SPARK Lab, is bridging the gap between humans and robots: perception. The group does theoretical and experimental research to expand a robot’s awareness of its environment, approaching human perception. While robots have improved in detecting and identifying objects, they still have a lot to learn about making higher-level sense of their surroundings.
The Quest for Higher-Level Perception
Carlone and his team are working to impart human-level perception to robots, enabling them to safely and seamlessly interact with people in their homes, workplaces, and other unstructured environments. They are developing perception and scene-understanding algorithms for various applications, including autonomous underground search-and-rescue vehicles, drones that can pick up and manipulate objects, and self-driving cars.
Expanding Horizons
Carlone was born and raised in Italy, where his parents, a teacher and a historian, instilled in him a love for math and history. He pursued engineering, and his passion for robotics grew during his undergraduate studies. He received his PhD in mechatronics and later became a postdoc at Georgia Tech, where he delved into coding and computer vision.
The Next Generation
Carlone’s research has focused on simultaneous localization and mapping (SLAM), a problem that robots face when generating and updating a map of their environment while tracking their location. His work helped crack open the field, which was previously considered solved.
Spatial AI
Today, Carlone’s group is developing ways to represent a robot’s surroundings beyond characterizing their geometric shape and semantics. They are using deep learning and large language models to create algorithms that enable robots to perceive their environment through a higher-level lens. This work fits into the emerging field of "spatial AI," which aims to enable robots to think and understand the world like humans do.
Conclusion
Carlone’s vision is to create robots that can perceive their environment in a way similar to humans. This will enable robots to interact with people more intuitively, making them more useful and effective in various settings. While there is still much work to be done, Carlone is confident that the future is bright for robots.
FAQs
Q: What is spatial AI?
A: Spatial AI is a field that aims to enable robots to think and understand the world like humans do, in ways that can be useful.
Q: How does Carlone’s research fit into this field?
A: Carlone’s research is developing ways to represent a robot’s surroundings beyond characterizing their geometric shape and semantics, using deep learning and large language models.
Q: What is the goal of Carlone’s work?
A: The goal is to create robots that can perceive their environment in a way similar to humans, enabling them to interact with people more intuitively and effectively.
