This third edition of Automatic Target Recognitionprovides a roadmap for breakthrough ATR designs with increased intelligence, performance, and autonomy. Clear distinctions are made between military problems and comparable commercial deep-learning problems. These considerations need to be understood by ATR engineers working in the defense industry as well as by their government customers.
A reference design is provided for a next-generation ATR that can continuously learn from and adapt to its environment. The convergence of diverse forms of data on a single platform supports new capabilities and improved performance. This third edition broadens the notion of ATR to multisensor fusion. Radical continuous-learning ATR architectures, better integration of data sources, well-packaged sensors, and low-power teraflop chips will enable transformative military designs.
