Drilling is not as modern as other industries and still has great potential for automatic control. Automatic drill pipe emission control system is mostly used for offshore drilling, and the derrick of the land drilling rig cannot be directly compatible with the mature drill pipe emission system of the offshore drilling rig because of its unique structure. Drill pipe emissions from land drilling are usually done by worker-assisted mechanical arm. In this kind of working environment, workers are labor intensive and prone to fatigue and safety cannot be effectively guaranteed. In order to reduce labor intensity and improve drilling efficiency, this paper presents an automatic control system of drill pipe emission based on machine vision. In the control process of the tripping operations, visual servo control is added, and the visual sensor is used to detect the centerline of the drill pipe in real time. After the system finding the deviation between the drill pipe and the mechanical arm, the posture of the arm is adjusted in real time to compensate for the deviation. When the claw mechanism is aligned with the centerline of the drill pipe, the drill pipe manipulator grabs the drill pipe to complete the emission task. In this paper, four methods for extracting the centerline are compared, and the linear regression method is the best. Keywords Machine vision • Mechanical arm • Drill pipe emission • Image processing List of symbols mm A unit of length in the metric system (millimeter) kN Common unit of mechanical calculation (1 kN = 1000 kg × 1 m/s 2) EX The abscissa of the intersection minus the abscissa of the center point of the lens (pixel) FPS Frame per second (Hz)

No abstract

No abstract

Recently, self-training and active learning have been proposed to alleviate this problem. Self-training can improve model accuracy with massive unlabeled data, but some pseudo labels containing noise would be generated with limited or imbalanced training data. And there will be suboptimal models if human guidance is absent. Active learning can select more effective data to intervene, while the model accuracy can not be improved because the massive unlabeled data are not used. And the probability of querying sub-optimal samples will increase when the domain difference is too large, increasing annotation cost. This paper proposes an iterative loop learning method combining Self-Training and Active Learning (STAL) for domain adaptive semantic segmentation. The method first uses self-training to learn massive unlabeled data to improve model accuracy and provide more accurate selection models for active learning. Secondly, combined with the sample selection strategy of active learning, manual intervention is used to correct the self-training learning. Iterative loop to achieve the best performance with minimal label cost. Extensive experiments show that our method establishes stateof-the-art performance on tasks of GTAV → Cityscapes, SYNTHIA → Cityscapes, improving by 4.9% mIoU and 5.2% mIoU, compared to the previous best method, respectively. Code will be available.

No abstract