In advanced technology nodes, benefits from scaling have become limited in terms of power, performance, and area (PPA), necessitating improvements through design-technology cooptimization (DTCO). While the standard-cell methodology is widely used in modern VLSI design, it inherently constrains the potential of DTCO due to its abstraction at the logic level, limiting optimization at the physical level. To bridge this gap, transistor-level design methodology is desired to break the abstraction of standard cells. Existing research has explored large-scale transistor placement, but a gap remains in developing a routing framework that efficiently addresses transistor-level routing challenges. This paper proposes a pioneering transistorlevel routing framework for large-scale transistor placements, along with an efficient CP-SAT (Constraint ProgrammingSATisfiability) formulation for routing in lower layers. Experimental results demonstrate that the proposed routing framework enables transistor-level designs to achieve significantly reduced total wirelength and high utilization rates for designs with hundreds to thousands of transistors, outperforming traditional standard-cell-based approaches in advanced technology nodes.

Object-goal navigation (Object-nav) entails searching, recognizing and navigating to a target object. Object-nav has been extensively studied by the Embodied-AI community, but most solutions are often restricted to considering static objects (e.g., television, fridge, etc.), We propose a modular framework for object-nav that is able to efficiently search indoor environments for not just static objects but also movable objects (e.g. fruits, glasses, phones, etc.) that frequently change their positions due to human intervention. Our contextual-bandit agent efficiently explores the environment by showing optimism in the face of uncertainty and learns a model of the likelihood of spotting different objects from each navigable location. The likelihoods are used as rewards in a weighted minimum latency solver to deduce a trajectory for the robot. We evaluate our algorithms in two simulated environments and a real-world setting, to demonstrate high sample efficiency and reliability.

OPL is a modeling language for mathematical programming and combinatorial optimization. It is the first language to combine high-level algebraic and set notations from mathematical modeling languages with a rich constraint language and the ability to specify search procedures and strategies that are the essence of constraint programming. This paper describes the facilities available in OPL to specify search procedures. It describes the abstractions of OPL to specify both the search tree (search) and how to explore it (strategies). The paper also illustrates how to use these high-level constructs to implement traditional search procedures in constraint programming and scheduling.