SBIR Phase I: High Performance Sense and Avoid

SBIR 第一阶段：高性能感知和避免

• 批准号: 1648560
• 负责人: Olivier Coenen
• 金额: $ 22.5万
• 依托单位: QELZAL CORPORATION
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2016
• 资助国家: 美国
• 起止时间: 2016-12-15 至 2017-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1648560&HistoricalAwards=false
• 关键词: SBIR; Phase; Performance; Sense; Avoid

The broader impact/commercial potential of this project is to develop and provide novel computation capabilities to the marketplace that mimic and apply the way our brain computes beyond deep learning systems and much closer to how the brain actually operates. Althoughthe technology is initially targeted for the commercial drone market, the technology can be applied to consumer and hobbyist drone market, self-driving cars and advanced driver assistance systems, autonomous navigation and guiding systems with obstacle avoidance forrobots, ballistics tracking and counter-drone capabilities for military and defense, and surveillance and counter-drone for public safety and security. This project has the potential to revolutionize robotic and machine vision by providing capabilities that simply do not exist today.This Small Business Innovation Research (SBIR) Phase I project investigates novel ways, algorithms and software implementations that leverage expertise in natural vision systems, deep learning and machine learning to make use of electro-optical sensors, whichrespond in new ways to form the basis of an Airborne Based Sense and Avoid (ABSAA) system. It leverages the benefits of bio-inspired computation, and investigates how to combine information from multiple sensors in a common representation. The project will study novel ways to achieve robust detection, segmentation, clustering, discrimination and classification with these novel sensors. It will also extend methods and algorithms for state estimation, tracking and prediction in the inherent sensor representation. The project will also address the real-time constraints and will attempt to leverage recent hardware implementations, which can provide a complete embedded system for ABSAA system that is low SWaP (size, weight and power) and in the long run, low cost as well because it has the potential to benefit from economy of scale. The theoretical and algorithmic advances generated by the project have the potential to affectmachine and robotic vision well beyond the project focused application to ABSAA.

该项目的更广泛的影响/商业潜力是为市场开发和提供新颖的计算能力，这些计算能力模仿并应用了我们的大脑超出深度学习系统的计算方式，并且更接近大脑的实际运作方式。尽管该技术最初是针对商业无人机市场的目标，但该技术可以应用于消费者和业余爱好者市场，自动驾驶汽车以及先进的驾驶员援助系统，自主导航和具有障碍型福克斯的指导系统，弹道跟踪和对公共安全和公共安全和柜台安全和柜台安全和柜台安全和柜台的功能。该项目具有通过提供今天根本不存在的功能来彻底改变机器人和机器视觉的潜力。本小型企业创新研究（SBIR）I期项目研究了在自然视觉系统中利用专业知识的新颖方式，算法和软件实现，深度学习和机器学习以使电动传感器的使用，以形成基于新型的基础（避免了基于空中）的（避免使用）的（避免使用）。它利用了生物启发的计算的好处，并研究了如何将来自多个传感器的信息组合在一起。该项目将研究与这些新型传感器实现强大检测，分割，聚类，歧视和分类的新颖方法。它还将扩展固有传感器表示中状态估计，跟踪和预测的方法和算法。该项目还将解决实时限制，并将试图利用最近的硬件实现，该实现可以为ABSAA系统提供完整的嵌入式系统，该系统是低交换（尺寸，重量和功率），从长远来看，也很低的成本，因为它有可能受益于规模经济。该项目产生的理论和算法进步具有影响力和机器人愿景的潜力，这远远超出了以项目为重点的ABSAA。