Abstract
When operating automated guided vehicles (AGVs) in a logistics or production environment, they are likely to face unknown obstacles, e.g. logistics containers or human personnel. In order to ensure a high productivity of the AGVs, instead of stopping, they are supposed to circumvent such obstacles if possible. At the same time, the jerk in the circumventing movement must be kept small to reduce wear on the drive system and to ensure that the AGV's load does not become disordered or damaged. Circumventing obstacles is done using local path planning. Local path planning aims to find a short-term path planning solution in an unknown environment.
However, common algorithms for local path planning, like vector field histogram or dynamic window, generate a steering command for the next interval without respect to smoothness of the path. If the generated paths are smoothed, this happens by statistically reducing jitter in the results. Such a smoothing method does not take the jerk into account.
This paper shows that paths of G2 differentiability are needed to reduce jerk on a path planning level. A path planning concept for the vector field histogram and for dynamic window and an evaluation method for this concept are shown.
Users
Please
log in to take part in the discussion (add own reviews or comments).
