查看更多>>摘要：News editors obtained the following quote from the background information suppli ed by the inventors: “For autonomous or semi-autonomous robotic devices to opera te autonomously or with minimal input and/or external control within an environm ent, methods for mapping the environment are helpful such that the robotic devic e may autonomously remain and operate within the environment. Methods for mappin g an environment have been previously proposed. For example, the collection and storage of a large amount of feature points from captured images of an environme nt wherein recognizable landmarks among the data may be identified and matched f or building and updating a map has been proposed. Such methods can require signi ficant processing power and memory due to the large amount of feature points ext racted from the captured images, their storage and sophisticated techniques used in creating the map. For example, some methods employ an EKF technique where th e pose of the robotic device and the position of features within the map of the environment are estimated and stored in a complete state vector while uncertaint ies in the estimates are stored in an error covariance matrix. The main drawback is the computational power required to process a large number of features havin g large total state vector and covariance matrix. Further, methods employing EKF can require accurate measurement noise covariance matrices a priori as inaccura te sensor statistics can lead to poor performance. Other methods of mapping an e nvironment use a distance sensor of the robotic device to measure distances from the distance sensor to objects within the environment while tracking the positi on of the robotic device. For example, a method has been proposed for constructi ng a map of the environment by rotating a distance sensor 360-degrees at a measu red rotational velocity while taking distance measurements to objects within the environment. While this method is simple, the method is limited as the mapping process of the environment relies on the distance sensor initially rotating 360- degrees. If the distance sensor is installed on a robotic device, for example, t he robotic device may rotate 360-degrees initially to finish mapping the environ ment before performing work. Another similar method provides that the robotic de vice may immediately translate and rotate while measuring distances to objects, allowing it perform work while simultaneously mapping. The method however uses E KF SLAM approach requiring significant processing power. Some mapping methods de scribe the construction of an occupancy map, where all points in the environment are tracked, including perimeters, empty spaces, and spaces beyond perimeters, and assigned a status, such as “occupied,” “unoccupied,” or “unknown.” This appr oach can have high computational costs. Other methods require the use of additio nal components for mapping, such as beacons, which must be placed within the env ironment. This is undesirable as additional components increase costs, take up s pace, and are unappealing to have within, for example, a consumer home.
NETL
NSTL