Search for: All records

Manipulating an articulated object requires perceiving its kinematic hierarchy: its parts, how each can move, and how those motions are coupled. Previous work has explored perception for kinematics, but none infers a complete kinematic hierarchy on never-before-seen object instances, without relying on a schema or template. We present a novel perception system that achieves this goal. Our system infers the moving parts of an object and the kinematic couplings that relate them. To infer parts, it uses a point cloud instance segmentation neural network and to infer kinematic hierarchies, it uses a graph neural network to predict the existence, direction, and type of edges (i.e. joints) that relate the inferred parts. We train these networks using simulated scans of synthetic 3D models. We evaluate our system on simulated scans of 3D objects, and we demonstrate a proof-of-concept use of our system to drive real-world robotic manipulation. 

The COVID-19 pandemic forced many colleges and universities to remain on a completely online or remote educational learning for more than a year; however, due to distraction, lack of motivation or engagement, and other internal/external pandemic contributing factors, learners could not pay attention 100% to the learning process. Additionally, given that transportation classes are very hands-on, students could not do the experiment from home due to limited resources available, thereby hampering all three phases of learner interactions. The limitation of the implementation of physical, hands-on laboratory exercises during the pandemic further exacerbated students’ actualization of the critical Accreditation Board for Engineering and Technology (ABET) outcomes in transportation: An ability to develop and conduct experiments or test hypotheses, analyze and interpret data and use scientific judgment to draw conclusions. Subsequently, this paper highlights the development and implementation of experiment centric pedagogy (ECP) home-based active learning experiments in three transportation courses: Introduction to Transportation Systems, Traffic Engineering, and Highway Engineering during the pandemic. Quantitative and qualitative student success key constructs data was collected in conjunction with the execution of classroom observation protocols that measure active learning in these transportation courses. The results reveal a significant difference between the pre, and post- tests of key constructs associated with student success, such as motivation, critical thinking, curiosity, collaboration, and metacognition. The results of the Classroom Observation Protocol for Undergraduate STEM (COPUS) show more active student engagement when ECP is implemented. 

Abstract We must be able to predict and mitigate against geomagnetically induced current (GIC) effects to minimize socio‐economic impacts. This study employs the space weather modeling framework (SWMF) to model the geomagnetic response over Fennoscandia to the September 7–8, 2017 event. Of key importance to this study is the effects of spatial resolution in terms of regional forecasts and improved GIC modeling results. Therefore, we ran the model at comparatively low, medium, and high spatial resolutions. The virtual magnetometers from each model run are compared with observations from the IMAGE magnetometer network across various latitudes and over regional‐scales. The virtual magnetometer data from the SWMF are coupled with a local ground conductivity model which is used to calculate the geoelectric field and estimate GICs in a Finnish natural gas pipeline. This investigation has lead to several important results in which higher resolution yielded: (1) more realistic amplitudes and timings of GICs, (2) higher amplitude geomagnetic disturbances across latitudes, and (3) increased regional variations in terms of differences between stations. Despite this, substorms remain a significant challenge to surface magnetic field prediction from global magnetohydrodynamic modeling. For example, in the presence of multiple large substorms, the associated large‐amplitude depressions were not captured, which caused the largest model‐data deviations. The results from this work are of key importance to both modelers and space weather operators. Particularly when the goal is to obtain improved regional forecasts of geomagnetic disturbances and/or more realistic estimates of the geoelectric field.