ABOUT
Closed-loop-Attention
Task 3.1
Hybrid Algorithms for Feature Extraction
Task Lead: Saibal Mukhopadhyay, Other PIs: Amit Ranjan Trivedi, Pamela Abshire
OBJECTIVE
This task will explore the computational building blocks for informative dimensionality reduction to reduce the data both before and after digitization, allowing the sensor to provide relevant information for the task at hand while making efficient use of available resources.
Task 3.2
Adaptive Feature Extraction
Task Lead: Justin Romberg, Other PIs: Stanley Chan, Mingoo Seok
OBJECTIVE
This task will develop techniques for adapting the feature extraction dynamically at the sensor as the environment changes. Adapting the features to the environment will allow dimensionality to be more efficient and more robust. Deliverable. Software implementation of adaptive feature extraction algorithms.
Task 3.3
Dynamic Power-Trust Scaling Closed-Loop Control
Task Lead: Mingoo Seok, Other PIs: Saibal Mukhopadhyay 2
OBJECTIVE
The goal of this task is to create low-latency yet power-efficient feedback and feedforward control systems for dynamic pixel power-trust scaling. Deliverable. Simulation models of control algorithms.
Task 3.4
Hardware Acceleration
Task Lead: Mingoo Seok, Other PIs: Saibal Mukhopadhyayeddx’
OBJECTIVE
This task develops the programmable and energy-efficient computing hardware platforms that execute the algorithms for trust estimation, adaptive feature extraction, RNN-based feedback control, and self-supervised feed-forward control. Deliverable. Design and test-chip measurement results of adaptation accelerators. constructs.
Task 3.5
Power-Trust Management Hardware
Task Lead: Alyosha Molnar, Other PIs: Cheng Huang
OBJECTIVE
The objective of this task is to develop circuit techniques to allow dynamic trade-offs of different sensing-relevant performance metrics against power consumption, for RF/mmWave radar as well as optical sensing such as LiDAR, and do so in ways that are compatible with rapid optimization algorithms and hardware. Deliverable. Circuit techniques and test-chip measurement results of pixel tuning circuits.