Gears are a vital component in many complex mechanical systems. In automotive systems, and in particular vehicle transmissions, we rely on them to function properly on different types of challenging environments and conditions. However, when a gear is manufactured with a defect, the gear’s integrity can become compromised and lead to catastrophic failure. The current inspection process used by an automotive gear manufacturer in Guelph, Ontario, requires human operators to visually inspect all gear produced. Yet, due to the quantity of gears manufactured, the diverse array of defects that can arise, the time requirements for inspection, and the reliance on the operator’s inspection ability, the system suffers from poor scalability, and defects can be missed during inspection. In this work, we propose a machine vision system for automating the inspection process for gears with damaged teeth defects. The implemented inspection system uses a faster R-CNN network to identify the defects, and combines domain knowledge to reduce the manual inspection of non-defective gears by 66%.

Harvesting operations in agriculture are labour-intensive tasks. Automated solutions can help alleviate some of the pressure faced by rising costs and labour shortage. Yet, these solutions are often difficult and expensive to develop. To enable the use of harvesting robots, machine vision must be able to detect and localize target objects in a cluttered scene. In this work, we focus on a subset of harvesting operations, namely, tomato harvesting in greenhouses, and investigate the impact that variations in dataset size, data collection process and other environmental conditions may have on the generalization ability of a Mask-RCNN model in detecting two objects critical to the harvesting task: tomatoes and stems. Our results show that when detecting stems from a perpendicular perspective, models trained using data from the same perspective are similar to one that combines both perpendicular and angled data. We also show larger changes in detection performance across different dataset sizes when evaluating images collected from an angled camera perspective, and overall larger differences in performance when illumination is the primary source of variation in the data. These findings can be used to help practitioners prioritize data collection and evaluation efforts, and lead to larger-scale harvesting dataset construction efforts.

Robotic grasping systems often rely on visual observations to drive the grasping process, where the robot must be able to detect and localize an object, extract features relevant to the task, and then combine this information to plan a manipulation strategy. But what happens when some of the most impactful features are not observed by the robot Without context on an objects center-of-mass, for example, a robot may make assumptions such as uniform density that do not hold, and which may in turn guide the robot into perceiving a suboptimal set of grasping configurations. In this work, we examine how having prior knowledge of an object's intrinsic properties influences the task of dense grasp affordance prediction. We investigate a simple, constrained grasping task where object properties heavily regulate the space of successful grasps, and further evaluate how learning is affected when generalizing across unseen weight configurations and unseen object shapes.

Manufacturing is an imperfect process that requires frequent checks and verifications to ensure products are being produced properly. In many cases, such as visual inspection, these checks can be automated to a certain degree. Incorporating advanced inspection techniques (i.e., via deep learning) into real-world inspection pipelines requires different mechanical, machine vision, and process-level considerations. In this work, we present an approach that builds upon prior work at an automotive gear facility located in Guelph, Ontario, which is looking to expand its defect detection capabilities. We outline a set of inspection-cell changes, which has led to full-gear surface scanning and inspection at a rate of every 7.5 s, and which is currently able to detect three common types of surface-level defects.

Fire blight is an infectious disease found in apple and pear orchards. While managing the disease is critical to maintaining orchard health, identifying symptoms early is a challenging task which requires trained expert personnel. This paper presents an inspection technique that targets individual symptoms via deep learning and density estimation. We evaluate the effects of including multi-spectral sensors in the model’s pipeline. Results show that adding near infrared (NIR) channels can help improve prediction performance and that density estimation can detect possible symptoms when severity is in the mid-high range.