Sharing information between connected and autonomous vehicles (CAVs) fundamentally improves the performance of collaborative object detection for self-driving. However, CAVs still have uncertainties on due to practical challenges, which will affect later modules in self-driving such as planning control. Hence, uncertainty quantification is crucial safety-critical systems CAVs. Our work first estimate detection. We propose a novel method, called Double- M Quantification, tailors moving block...

Object detection and multiple object tracking (MOT) are essential components of self-driving systems. Accurate uncertainty quantification both critical for onboard modules, such as perception, prediction, planning, to improve the safety robustness autonomous vehicles. Collaborative (COD) has been proposed accuracy reduce by leveraging viewpoints agents. However, little attention paid how leverage from COD enhance MOT performance. In this letter, first attempt address challenge, we design an...

With the development of sensing and communication technologies in networked cyber-physical systems (CPSs), multi-agent reinforcement learning (MARL)-based methodologies are integrated into control process physical demonstrate prominent performance a wide array CPS domains, such as connected autonomous vehicles (CAVs). However, it remains challenging to mathematically characterize improvement CAVs with cooperation capability. When each individual vehicle is originally self-interest, we can...

Various methods for Multi-Agent Reinforcement Learning (MARL) have been developed with the assumption that agents' policies are based on accurate state information. However, learned through Deep (DRL) susceptible to adversarial perturbation attacks. In this work, we propose a State-Adversarial Markov Game (SAMG) and make first attempt investigate different solution concepts of MARL under uncertainties. Our analysis shows commonly used optimal agent policy robust Nash equilibrium do not...

In real-world multi-agent reinforcement learning (MARL) applications, agents may not have perfect state information (e.g., due to inaccurate measurement or malicious attacks), which challenges the robustness of agents' policies. Though is getting important in MARL deployment, little prior work has studied uncertainties MARL, neither problem formulation nor algorithm design. Motivated by this issue and lack corresponding studies, we study with uncertainty work. We provide first attempt...

Approximate computing is an emerging energy-efficient paradigm for error-resilient applications. logic synthesis (ALS) important field of it. To improve the existing ALS flows, one key issue to derive a more accurate and efficient batch error estimation technique all approximate transformations under consideration. In this work, we propose novel method based on Monte Carlo simulation local change propagation. It generally applicable any statistical measurement such as rate average magnitude....

In the realm of autonomous vehicle (AV) perception, comprehending 3D scenes is paramount for tasks such as planning and mapping. Semantic scene completion (SSC) aims to infer geometry semantics from limited observations. While camera-based SSC has gained popularity due affordability rich visual cues, existing methods often neglect inherent uncertainty in models. To address this, we propose an uncertainty-aware semantic method ($\alpha$-SSC). Our approach includes propagation framework depth...

Object detection and multiple object tracking (MOT) are essential components of self-driving systems. Accurate uncertainty quantification both critical for onboard modules, such as perception, prediction, planning, to improve the safety robustness autonomous vehicles. Collaborative (COD) has been proposed accuracy reduce by leveraging viewpoints agents. However, little attention paid how leverage from COD enhance MOT performance. In this paper, first attempt address challenge, we design an...

With the development of sensing and communication technologies in networked cyber-physical systems (CPSs), multi-agent reinforcement learning (MARL)-based methodologies are integrated into control process physical demonstrate prominent performance a wide array CPS domains, such as connected autonomous vehicles (CAVs). However, it remains challenging to mathematically characterize improvement CAVs with cooperation capability. When each individual vehicle is originally self-interest, we can...

Sharing information between connected and autonomous vehicles (CAVs) fundamentally improves the performance of collaborative object detection for self-driving. However, CAVs still have uncertainties on due to practical challenges, which will affect later modules in self-driving such as planning control. Hence, uncertainty quantification is crucial safety-critical systems CAVs. Our work first estimate detection. We propose a novel method, called Double-M Quantification, tailors moving block...