Indoor wireless propagation is a prominent factor for the success of remote sensing applications, such as smart homes, inventory management, and emergency support. In scenarios where motion is involved, safely navigating and localizing in dynamic indoor environments comprising of various stationary and mobile obstacles but with limited GPS coverage is of valid concern, especially for vulnerable individuals. One solution to prevent harm while moving at greater speeds, such as concussions for football players or collision avoidance for autonomous bots, is to make use of active devices that are capable of sensing the surroundings, communicating and collaborating with other devices, and equip them with decision-making algorithms for alerting about the impending accidents. This proposal provides an insight into the work done and planned in the domain for (a) predicting impending collisions and (b) locating moving entities, utilizing the principles of indoor wireless radio propagation, along with inertial sensing and doppler.

 Specifically, for the former goal, we focus on designing and implementing a smart helmet system. One implementation is realized with the help of packet-based, UWB radio communication protocol to build a pairwise range-based, real-time, decentralized collision prediction algorithm called 'Autonomous Collision Estimation for Dynamic Motion' or 'ACED' that doesn't require calibration or known-location/ beacon node placement. We provide a prototype to make a case for active UWB wearable tags on moving objects as they are low cost, low power, robust, compact, and thus ideal for real-world implementations on spaceconstrained equipment such as headgears and drones. We extend this problem statement to sense passive obstacles with the help of a COTS, mm-wave, FMCW Radar and build a deep learning-based collision prediction system, called 'COPR', abbreviated for 'Collision prediction using Radars' and provide a real-world implementation of it using a 24.5 GHz radar system. For the latter goal of locating moving objects, we propose a novel technique of using dopplerspread, which is a result of multipath propagation, a prominent challenge that is faced in indoor environment, and building a deep-learning based location estimation algorithm. Overall, we demonstrate and investigate techniques that use affordable wireless technologies to sense moving objects in order to assist in their health and safety.

Indoor wireless propagation is a prominent factor for the success of remote sensing applications, such as smart homes, inventory management, and emergency support. In scenarios where motion is involved, safely navigating and localizing in dynamic indoor environments comprising of various stationary and mobile obstacles but with limited GPS coverage is of valid concern, especially for vulnerable individuals. One solution to prevent harm while moving at greater speeds, such as concussions for football players or collision avoidance for autonomous bots, is to make use of active devices that are capable of sensing the surroundings, communicating and collaborating with other devices, and equip them with decision-making algorithms for alerting about the impending accidents. This proposal provides an insight into the work done and planned in the domain for (a) predicting impending collisions and (b) locating moving entities, utilizing the principles of indoor wireless radio propagation, along with inertial sensing and doppler.

 Specifically, for the former goal, we focus on designing and implementing a smart helmet system. One implementation is realized with the help of packet-based, UWB radio communication protocol to build a pairwise range-based, real-time, decentralized collision prediction algorithm called 'Autonomous Collision Estimation for Dynamic Motion' or 'ACED' that doesn't require calibration or known-location/ beacon node placement. We provide a prototype to make a case for active UWB wearable tags on moving objects as they are low cost, low power, robust, compact, and thus ideal for real-world implementations on spaceconstrained equipment such as headgears and drones. We extend this problem statement to sense passive obstacles with the help of a COTS, mm-wave, FMCW Radar and build a deep learning-based collision prediction system, called 'COPR', abbreviated for 'Collision prediction using Radars' and provide a real-world implementation of it using a 24.5 GHz radar system. For the latter goal of locating moving objects, we propose a novel technique of using dopplerspread, which is a result of multipath propagation, a prominent challenge that is faced in indoor environment, and building a deep-learning based location estimation algorithm. Overall, we demonstrate and investigate techniques that use affordable wireless technologies to sense moving objects in order to assist in their health and safety.