Learning Efficient Policies for Picking Entangled Wire Harnesses: An Approach to Industrial Bin Picking

Abstract

Industrial bin picking for tangled-prone objects requires the robot to either pick up untangled objects or perform separation manipulation when the bin contains no isolated objects. The robot must be able to flexibly perform appropriate actions based on the current observation. It is challenging due to high occlusion in the clutter, elusive entanglement phenomena, and the need for skilled manipulation planning. In this paper, we propose an autonomous, effective and general approach for picking up tangled-prone objects for industrial bin picking. First, we learn PickNet -a network that maps the visual observation to pixel-wise possibilities of picking isolated objects or separating tangled objects and infers the corresponding grasp. Then, we propose two effective separation strategies: Dropping the entangled objects into a buffer bin to reduce the degree of entanglement; Pulling to separate the entangled objects in the buffer bin planned by PullNet -a network that predicts position and direction for pulling from visual input. To efficiently collect data for training PickNet and PullNet, we embrace the self-supervised learning paradigm using an algorithmic supervisor in a physics simulator. Realworld experiments show that our policy can dexterously pick up tangled-prone objects with success rates of 90%. We further demonstrate the generalization of our policy by picking a set of unseen objects. Supplementary material, code, and videos can be found at https://xinyiz0931.github.io/tangle.

Bin picking: a review of recent developments

“…The system (Figure 3) has been tested with PVC tubes and rubber vehicle radiator hoses (Figure 4), and experiments showed that the robot was able to pick a single item on the first try with success rates of 99% and 93%, respectively. A group from Osaka University has addressed the problem of untangling wiring harnesses (Zhang et al , 2023a). Training or collecting data in simulation is challenging due to the difficulties in modelling the combination of deformable and rigid components in the harnesses, and the group has developed a technique to grasp and extract the component by following a circle-like trajectory until it is untangled.…”

“…In a series of experiments, the system picked and separated tangled harnesses with a success rate of 84.6%. In subsequent work, the group deployed a dual-arm robot that can grasp, extract and disentangle wiring harnesses from a dense clutter using dynamic manipulation (Zhang et al , 2023b). The robot can swing to discard dynamically the entangled objects and then re-grasp to adjust the pose.…”

CAD-Based Robot Programming Solution for Wire Harness Manufacturing in Aeronautic Sector

“…Continuing with the use of collaborative robots, Navas et al [17] suggest the use of cobots to reduce ergonomic risks for wire harness assembly workers. Dual-arm approaches can also be found in the literature like the work presented by Zhang et al [18] where a closed-loop bin-picking system focused on entangled wire harnesses is presented. Lv et al [19] also present a dynamic model to control deformable linear objects (DLO) for single and dual-arm configurations.…”

Fine-grained Load Balancing with Traffic-aware Rerouting in DataCenter Networks

“…Deformable object manipulation has been partially investigated by predicting robot actions through the utilization of visual data [18], [19]. Cable grasping with complex geometries in cluttered environment, however, remains relatively unexplored [20], which will be the focus of this paper. Poor visual prediction still limits performance of cable grasping in heavily occluded clutter [20].…”

“…Cable grasping with complex geometries in cluttered environment, however, remains relatively unexplored [20], which will be the focus of this paper. Poor visual prediction still limits performance of cable grasping in heavily occluded clutter [20]. Motivated by this, we learn and generalize robotic policy with collision awareness in simulation using domain randomization.…”