Despite decades of research, existing navigation systems still face real-world challenges when deployed in the wild, e.g., in cluttered home environments or in human-occupied public spaces. To address this, we present a new class of implicit control policies combining the benefits of imitation learning with the robust handling of system constraints from Model Predictive Control (MPC). Our approach, called Performer-MPC, uses a learned cost function parameterized by vision context embeddings provided by Performers -- a low-rank implicit-attention Transformer. We jointly train the cost function and construct the controller relying on it, effectively solving end-to-end the corresponding bi-level optimization problem. We show that the resulting policy improves standard MPC performance by leveraging a few expert demonstrations of the desired navigation behavior in different challenging real-world scenarios. Compared with a standard MPC policy, Performer-MPC achieves >40% better goal reached in cluttered environments and >65% better on social metrics when navigating around humans.

The very idea of a driverless vehicle rolling around on the streets seems incredible. And yet, we may be close to seeing such vehicles on the road around the world, thanks to artificial intelligence (AI), among other driving forces. In the recent past, there have been some amazing advances in autonomous vehicle technology which indicate the dream is inching toward fruition. It seems that the framework of autonomous vehicles has been almost finalized. Subject to legal and administrative approvals, driverless vehicles will be a common sight on the roads soon.

Artificial intelligence that allows self-driving cars to detect people and objects hidden around blind corners has been developed by researchers at MIT. The imaging system--dubbed CornerCameras--was built by AI researchers at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) for seeing around obstructions using standard camera technology. Using information about light reflections, MIT's artificial intelligence system is able to measure the speed and trajectory of hidden objects in real time using footage from smartphone cameras. "The technology has a range of applications, from firefighters finding people in burning buildings to self-driving cars detecting pedestrians in their blind spots," an MIT spokesperson tells Newsweek. "What's impressive is that this approach works using footage from a smartphone camera, such as an iPhone 8." The artificial intelligence system can be used on footage filmed with a smartphone.