b'87He used similar image-based predictions in his first job at Disney Research, where he created systems to track basketball players, predict their behavior, and follow the action in real time in order to film live games automat-ically with robotic cameras. Later, he developed visual recognitionsoftwaretoanalyzeemotionalresponses to new programming based on viewers facial reactions Peter was eight years oldwhen the water pipeand body movements.in his familys cabin burst. They had just arrived in theEight years into his stint at Disney, Peter was offered countryside a few hours outside Toronto, expecting frig- an opportunity to combine all of his passionsartificial id temperatures and hoping for a few days respite fromintelligence,behavioralpredictions,buildingphysical the city. But the mixture of frozen water and underusedhardware, writing software for computer visionas the pipes made the Carr familys arrival considerably less re- head of autonomy for Argo AI. Today, he oversees all sen-laxing. With no plumbers available in their neck of thesor-based analysis, such as detection and classification, Canadian woods, they realized theyd need to fix the pipethat helps the self-driving system understand whats in themselves if they wanted running water anytime soon.its environment, as well as all planning and action soft-And so Peter, the youngest and smallest of the threeware that determines how to navigate from while inter-Carr kids, was tasked with squeezing into the crawlspaceacting with every object within that environment. (Any under the cabin. He was more than happy to offer his ser- time an Argo vehicle distinguishes between a bus and a vices. An avid LEGO user, Peter always had a knack fortractor-trailer, or reacts to a vehicle running a red light, fixing, building, and tinkering. A burst pipe was just an- thats Peters domain.)other problem for him to solve. It was his first real tastePeter still has ample opportunity to work on applied of applied engineering.scienceprojectsatleastofthedomesticvariety.Al-When he arrived at Queens University in Kingston,though its been decades since hes had to repair a burst Ontario, a decade later, Peter decided to major in en- water pipe, his two children, ages four and seven (and gineeringphysics.Inacourseonmedicalphysics,heavid LEGO fanatics themselves) are constantly providing learned how CT scans are generated from X-ray data tohim things in need of fixing. Hes still happy to roll up his help make diagnoses. This was the first time he realizedsleeves, climb into whatever nook and cranny necessary, the power of computer software applied to images, andand engineer a solution.decided to shift into the field of computer vision. He followed this interest to a PhD in Canberra, Aus-tralia, where he learned how to use cameras to detect, monitor, and predict athletes behavior, offering tactical support to the Australian Olympic field hockey team. THE BEHAVIORALISTFullBook_Mar24.indb 87 4/25/21 6:43 PM'