Team PLAID’s objective is to develop an autonomous robot that can pick and stow various objects in a warehouse environment.
The system should be capable of identifying, localizing, grasping, and transporting items from a shelf unit to one of three desired storage totes. This system design is based off of work from the Fall 2016 semester, and incorporates an actuated framing and shelf system, as well as multiple vision sensors, in conjunction with a robotic arm which utilizes an end effector and grasper. Due to the project’s reliance on the continual rollout of information for this year’s Amazon Robotics Challenge, the schedule for tasks have been chosen such that as much of the system can be completed in separate identifiable stages as possible with present information in mind.
To meet the functional and nonfunctional requirements of the picking challenge, the team focused on accurate item classification, efficient motion planning, robust grasping and customized storage system. These are also the bottlenecks fully automate warehouse material handling jobs. Amazon warehouse is currently semi-automated by having Kiva Pods that bring shelves to people for picking items on the work orders. To automate the picking part, the robotic picking system needs to handle the issues of accuracy and speed for picking wide range of items in limited time.
The final goal is to establish and implement the full working system with all hardware installed and controlled by April 26th 2017, in conjunction with the Spring Validation Experiment.Amazon Robotics Challenge rules for picking task was the major design requirement for the system. The SVE goal used the same 15 min time frame, known itemset, workcell layout, hardware design restriction, work order file specified by 2017 ARC. The competition will have known items and unknown items which would be released 30 mins before the competition and was out of scope for the MRSD project.