Search for: All records

Abstract Precise segmentation of irregular and densely arranged components is essential for robotic disassembly and material recovery in electronic waste (e-waste) recycling. This study evaluates the impact of model architecture and scale on segmentation performance by comparing SAM2, a transformer-based vision model, with the lightweight YOLOv8 network. Both models were trained and tested on a newly collected dataset of 1,456 annotated RGB images of laptop components including logic boards, heat sinks, and fans, captured under varying illumination and orientation conditions. Data augmentation techniques, such as random rotation (±15°), flipping, and cropping, were applied to improve model robustness. YOLOv8 achieved higher segmentation accuracy (mAP50 = 98.8%, mAP50-95 = 85%) and stronger boundary precision than SAM2 (mAP50 = 8.4%). SAM2 demonstrated flexibility in representing diverse object structures but often produced overlapping masks and inconsistent contours. These findings show that large pre-trained models require task-specific optimization for industrial applications. The resulting dataset and benchmarking framework provide a foundation for developing scalable vision algorithms for robotic e-waste disassembly and circular manufacturing systems. 

Metalloenzyme superfamilies are typically defined by their protein scaffolds and active sites. Owing to the high tunability of protein structures, members of a single superfamily can catalyze diverse reactions with the same metallocofactor. Some superfamilies, such as Amidohydrolase-related dinuclear oxygenases (ARO), display further versatility by utilizing multiple metallocofactors. We have shown that certain AROs catalyze monooxygenation reactions with diiron, dimanganese and/or mixed manganese-iron cofactors, but the molecular factors governing the selection of a particular cofactor remain unknown and the extent of this superfamily in biology is unclear. Here, we report bioinformatic analyses that expand the ARO superfamily to approximately 17,000 unique UniProt sequences, far exceeding the number of previously characterized enzymes. Through the integration of structural, spectroscopic and thermodynamic analyses of representative proteins with a bioinformatic pipeline that identifies key secondary- and tertiary-sphere residues, we can predict in silico the metal preference for the majority of reported ARO sequences. These annotations were validated via the characterization of multiple new AROs, including ones implicated in key oxidative steps of natural product biosyntheses. This study establishes the key structure-function relationships governing metal preferences in AROs and highlights their vastly underappreciated role in myriad biological processes. 

Most existing robotic surgery systems adopt a human-in-the-loop paradigm, often with the surgeon directly teleoperating the robotic system. Adding intelligence to these robots would enable higher-level control, such as supervised autonomy or even full autonomy. However, artificial intelligence (AI) requires large amounts of training data, which is currently lacking. This work proposes SurgSync, a multi-modal data collection framework with offline and online synchronization to support training and real-time inference, respectively. The framework is implemented on a da Vinci Research Kit (dVRK) and introduces (1) dual-mode (online/offline-matching) synchronized recorders, (2) a modern stereo endoscope to achieve image quality on par with clinical systems, and (3) additional sensors such as a side-view camera and a novel capacitive contact sensor to provide ground truth contact data. The framework also incorporates a post-processing toolbox for tasks such as depth estimation, optical flow, and a practical kinematic reprojection method using Gaussian heatmap. User studies with participants of varying skill levels are performed with ex-vivo tissue to provide clinically realistic data, and a network for surgical skill assessment is employed to demonstrate utilization of the collected data. Through the user study experiments, we obtained a dataset of 214 validated instances across multiple canonical training tasks. All software and data are available at surgsync.github.io. 

Abstract Guaianolide sesquiterpene natural products are of great biological and synthetic importance. Among them, phaeocaulisin A with a unique tetracyclic skeleton was isolated in 2013 and found to exhibit anticancer cell proliferation activity and anti-inflammatory activity. The first total synthesis of ( − )-phaeocaulisin A was achieved by Procter and coworkers in 2022. Two years later, we reported a total synthesis of ( ± )-phaeocaulisin A. The key steps of our total synthesis include a novel palladium-catalyzed cyclopropanol ring-opening carbonylative esterification to access a key γ-ketoester, a regio- and stereo-selective aldol cyclization to build the 7-membered carbocycle, and a cascade ketalization/lactonization to construct the tetracyclic skeleton. This Synpact article reflects our journey toward ( ± )-phaeocaulisin A, which also led to the discovery of an analogue with potent anticancer activity against triple negative or HER2+ breast cancer.