MIT engineers use new methods to train delivery robots to navigate more accurately to customers’ doorsteps

Some delivery robots now offer customers faster and more convenient means of transportation, but they have also forced technology companies to figure out the best way to safely guide robots to customers’ doorsteps, foreign media reported. Mapping neighborhoods in detail is a solution, but over time it becomes less reliable. The layout of the front yard can change (due to holidays, weather, renovations or new owners), which can also compromise customer privacy.

MIT工程师使用新方法训练送货机器人更准确地导航到客户家门口

So what better option? The Massachusetts Institute of Technology (MIT) may have found a solution to the problem in partnership with Ford. Engineers at the Massachusetts Institute of Technology recently developed a preliminary navigation method that uses environmental cues to help delivery robots instantly plan their route to their destination.

This method relies primarily on machine learning and artificial intelligence algorithms, rather than specific or semi-specific GPS coordinates.

You can see the running technology in the video above (in a digital environment in some way), but in short, MIT robots are taught how to combine information to form a complete path. For example, if you “tell” an MIT robot to deliver a package to someone’s garage, it might look for a nearby sidewalk that usually leads to a driveway and is almost certainly connected to the garage.

Researchers at the Massachusetts Institute of Technology believe this approach is more effective than older systems, which rely more on exploration (and therefore take more time). The Massachusetts Institute of Technology trained the robot son with satellite imagery taken from Bing Maps. The team reportedly assigned “semantic labels” to “context ugfeatures” in the images, such as gray for the front door, blue for the driveway, green for the hedge, and so on.

Add a Comment

Your email address will not be published. Required fields are marked *