Search for: All records

This paper presents an attention-based, deep learning framework that converts robot camera frames with dynamic content into static frames to more easily apply simultaneous localization and mapping (SLAM) algorithms. The vast majority of SLAM methods have difficulty in the presence of dynamic objects appearing in the environment and occluding the area being captured by the camera. Despite past attempts to deal with dynamic objects, challenges remain to reconstruct large, occluded areas with complex backgrounds. Our proposed Dynamic-GAN framework employs a generative adversarial network to remove dynamic objects from a scene and inpaint a static image free of dynamic objects. The Dynamic-GAN framework utilizes spatial-temporal transformers, and a novel spatial-temporal loss function. The evaluation of Dynamic-GAN was comprehensively conducted both quantitatively and qualitatively by testing it on benchmark datasets, and on a mobile robot in indoor navigation environments. As people appeared dynamically in close proximity to the robot, results showed that large, feature-rich occluded areas can be accurately reconstructed with our attention-based deep learning framework for dynamic object removal. Through experiments we demonstrate that our proposed algorithm has up to 25% better performance on average as compared to the standard benchmark algorithms. 

Direct write Inkjet Printing is a versatile additive manufacturing technology that allows for the fabrication of multiscale structures with dimensions spanning from nano to cm scale. This is made possible due to the development of novel dispensing tools, enabling controlled and precise deposition of fluid with a wide range of viscosities (1 – 50 000 mPas) in nanoliter volumes. As a result, Inkjet printing has been recognized as a potential low-cost alternative for several established manufacturing methods, including cleanroom fabrication. In this paper, we present a characterization study of PEDOT: PSS polymer ink deposition printing process realized with the help of an automated, custom Direct Write Inkjet system. PEDOT: PSS is a highly conductive ink that possesses good film forming capabilities. Applications thus include printing thin films on flexible substrates for tactile (touch) sensors. We applied the Taguchi Design of Experiment (DOE) method to produce the optimal set of PEDOT:PSS ink dispensing parameters, to study their influence on the resulting ink droplet diameter. We experimentally determined that the desired outcome of a printed thin film with minimum thickness is directly related to 1) the minimum volume of dispensed fluid and 2) the presence of a preprocessing step, namely air plasma treatment of the Kapton substrate. Results show that an ink deposit with a minimum diameter of 482 μm, and a thin film with approximately 300 nm thickness were produced with good repeatability.

 

The placement of SMD components is usually performed with Cartesian type robots, a task known as pick-and-place (P&P). Small Selective Compliance Articulated Robot Arm (SCARA) robots are also growing in popularity for this use because of their quick and accurate performance. This paper describes the use of the Lean Robotic Micromanufacturing (LRM) framework applied on a large, 10kg payload, industrial SCARA robot for PCB assembly. The LRM framework guided the precision evaluation of the PCB assembly process and provided a prediction of the placement precision and yield. We experimentally evaluated the repeatability of the system, as well as the resulting collective errors during the assembly. Results confirm that the P&P task can achieve the required assembly tolerance of 200 microns without employing closed-loop visual servoing, therefore considerably decreasing the system complexity and assembly time.