Most last-mile autonomous delivery robots today meet customers at the curb since mapping has been a barrier for delivery bots to go all the way to the front door. As mapping an entire neighborhood with the level of specificity required to enable front-door delivery would be very time and resource consuming, this solution could possibly reduce many of the challenges associated with this task.
With this new method, the delivery bot is equipped with a navigation method that lets the robot process cues in its surroundings on the fly to figure out the location of the front door. Basically, the robot identifies objects in its surroundings and labels them, using a “cost-to-go” map. The map uses data from training maps to color-code the surroundings into a heat map where it can determine which parts are more likely to be close to a “front door” and which are not. It then charts the most efficient path to the door based on that information. This method is a variation of SLAM (simultaneous localization and mapping).
Real-time localization, navigation, and mapping are some of the major issues that autonomous delivery robots face. By increasing the localization autonomy, this method has the possibility to create huge benefits for the companies that want to use delivery bots. Also, many of the challenges that delivery bots face are related to edge-cases (the scenarios that come up when the bot is being used in a live environment). This method could be a way to also address that issue.
Written by Mahdere DW Amanuel, RISE Viktoria.