Last week, attendees of the 2017 International Conference on Intelligent Robots and Systems (IROS) toured Motion Metrics as part of the five-day event. This annual conference is organized by the Institute for Electrical and Electronics Engineers (IEEE).
Established in 1988, IROS is a highly-regarded international robotics conference for researchers and professionals from academia and industry. The event was held in Vancouver this year and serves as a collaborative forum for the international robotics research community. Besides offering tours of high-tech robotics companies like Motion Metrics, IROS conferences host workshops, technical sessions, panel discussions, and multimedia presentations from experts in the field of robotics and automation. Motion Metrics was recommended for inclusion in the program by the world-class Collaborative Advanced Robotics and Intelligent Systems (CARIS) Laboratory at UBC, led by Professor Elizabeth Croft.
IROS has inspired innovation at Motion Metrics since day one, when President & CEO Dr. Tafazoli first presented his paper “Impedance control of a tele-operated hydraulic excavator”, co-authored with UBC researchers and professors, at its 8th annual conference. Viewing a mining shovel as an industrial robotic arm with four degrees of freedom (4 DOF) used in the field, then addressing sensor-based monitoring and closed-loop control challenges, prompted the genesis of Motion Metrics two years later. Today, intelligent control systems are employed in various mining processes, and Motion Metrics continues to draw motivation from such high-profile conferences as IEEE IROS as we prepare to enter the arena of autonomous mining.
IROS tour members included graduate students, professors, and industry professionals from around the world, and were especially interested in the deep learning-based object recognition technologies that power our flagship products. Historically, these products have relied on traditional computer vision techniques wherein commonly known feature descriptors for object detection work alongside classic machine learning algorithms; but when the Motion Metrics research and development team felt that they had reached the performance limits of tradition computer vision, they turned to artificial neural networks (ANNs) to improve empirical success. Guided by an optimization algorithm, deep neural networks rapidly learn a bank of features based on the data (training images) provided; the result is a more scalable system that approaches human-level accuracy.
Motion Metrics employees work alongside industry experts to bring the latest technologies in artificial intelligence, embedded computing, cloud computing, and stereo imaging to mining. To learn more about opportunities with Motion Metrics, visit our career page.