Andrea Censi’s research while affiliated with ETH Zurich and other places

Transportation network design often involves multiple stakeholders with diverse priorities. We consider a system with a hierarchical multi-agent structure, featuring self-optimized subnetwork operators at the lower level and a central organization at the upper level. Independent regional planning can lead to inefficiencies due to the lack of coordination, hindering interregional travel and cross-border infrastructure development, while centralized methods may struggle to align local interests and can be impractical to implement. To support decision making for such a system, we introduce an iterative VCG-based mechanism for multi-regional network design that fosters cooperation among subnetwork operators. By leveraging the Vickery-Clarke-Groves (VCG) mechanism, the framework determines collective investment decisions and the necessary payments from both operators and the central organization to achieve efficient outcomes. A case study on the European Railway System validates the effectiveness of the proposed method, demonstrating significant improvements in overall network performance through enhanced cross-region cooperation.

This paper discusses the integration challenges and strategies for designing mobile robots, by focusing on the task-driven, optimal selection of hardware and software to balance safety, efficiency, and minimal usage of resources such as costs, energy, computational requirements, and weight. We emphasize the interplay between perception and motion planning in decision-making by introducing the concept of occupancy queries to quantify the perception requirements for sampling-based motion planners. Sensor and algorithm performance are evaluated using False Negative Rates (FPR) and False Positive Rates (FPR) across various factors such as geometric relationships, object properties, sensor resolution, and environmental conditions. By integrating perception requirements with perception performance, an Integer Linear Programming (ILP) approach is proposed for efficient sensor and algorithm selection and placement. This forms the basis for a co-design optimization that includes the robot body, motion planner, perception pipeline, and computing unit. We refer to this framework for solving the co-design problem of mobile robots as CODEI, short for Co-design of Embodied Intelligence. A case study on developing an Autonomous Vehicle (AV) for urban scenarios provides actionable information for designers, and shows that complex tasks escalate resource demands, with task performance affecting choices of the autonomy stack. The study demonstrates that resource prioritization influences sensor choice: cameras are preferred for cost-effective and lightweight designs, while lidar sensors are chosen for better energy and computational efficiency.

This paper discusses the integration challenges and strategies for designing mobile robots, by focusing on the taskdriven, optimal selection of hardware and software to balance safety, efficiency, and minimal usage of resources such as costs, energy, computational requirements, and weight. We emphasize the interplay between perception and motion planning in decisionmaking by introducing the concept of occupancy queries to quantify the perception requirements for sampling-based motion planners. Sensor and algorithm performance are evaluated using False Negative Rate (FNR) and False Positive Rate (FPR) across various factors such as geometric relationships, object properties, sensor resolution, and environmental conditions. By integrating perception requirements with perception performance, an Integer Linear Programming (ILP) approach is proposed for efficient sensor and algorithm selection and placement. This forms the basis for a co-design optimization that includes the robot body, motion planner, perception pipeline, and computing unit. We refer to this framework for solving the co-design problem of mobile robots as CODEI, short for Co-design of Embodied Intelligence. A case study on developing an Autonomous Vehicle (AV) for urban scenarios provides actionable information for designers, and shows that complex tasks escalate resource demands, with task performance affecting choices of the autonomy stack. The study demonstrates that resource prioritization influences sensor choice: cameras are preferred for cost-effective and lightweight designs, while lidar sensors are chosen for better energy and computational efficiency.

The large-scale allocation of public resources (e.g., transportation, energy) is among the core challenges of future Cyber-Physical-Human Systems (CPHS). In order to guarantee that these systems are efficient and fair, recent works have investigated non-monetary resource allocation schemes, including schemes that employ karma. Karma is a non-tradable token that flows from users gaining resources to users yielding resources. Thus far karma-based solutions considered the allocation of a single public resource, however, modern CPHS are complex as they involve the allocation of multiple coupled resources. For example, a user might want to trade-off fast travel on highways for convenient parking in the city center, and different users could have heterogeneous preferences for such coupled resources. In this paper, we explore how to optimally combine multiple karma economies for coupled resource allocations, using two mechanism-design instruments: (non-uniform) karma redistribution; and (non-unit) exchange rates. We first extend the existing Dynamic Population Game (DPG) model that predicts the Stationary Nash Equilibrium (SNE) of the multi-karma economies. Then, in a numerical case study, we demonstrate that the design of redistribution significantly affects the coupled resource allocations, while non-unit exchange rates play a minor role. To assess the allocation outcomes under user heterogeneity, we adopt Nash welfare as our social welfare function, since it makes no interpersonal comparisons and it is axiomatically rooted in social choice theory. Our findings suggest that the simplest mechanism design, that is, uniform redistribution with unit exchange rates, also attains maximum social welfare.

Network-based complex systems are inherently interconnected, with the design and performance of subnetworks being interdependent. However, the decisions of self-interested operators may lead to suboptimal outcomes for users. In this paper, we consider the question of what cooperative mechanisms can benefit both operators and users simultaneously. We address this question in a game theoretical setting, integrating both non-cooperative and cooperative game theory. During the non-cooperative stage, subnetwork decision-makers strategically design their local networks. In the cooperative stage, the co-investment mechanism and the payoff-sharing mechanism are developed to enlarge collective benefits and fairly distribute them. A case study of the Sioux Falls network is conducted to demonstrate the efficiency of the proposed framework. The impact of this interactive network design on environmental sustainability, social welfare and economic efficiency is evaluated, along with an examination of scenarios involving regions with heterogeneous characteristics.

This paper proposes a nonmonetary traffic demand management scheme, named CARMA, as a fair solution to the morning commute congestion. We consider heterogeneous commuters traveling through a single bottleneck that differ in both the desired arrival time and value of time (VOT). We consider a generalized notion of VOT by allowing it to vary dynamically on each day (e.g., according to trip purpose and urgency) rather than being a static characteristic of each individual. In our CARMA scheme, the bottleneck is divided into a fast lane that is kept in free flow and a slow lane that is subject to congestion. We introduce a nontradable mobility credit, named karma, that is used by commuters to bid for access to the fast lane. Commuters who get outbid or do not participate in the CARMA scheme instead use the slow lane. At the end of each day, karma collected from the bidders is redistributed, and the process repeats day by day. We model the collective commuter behaviors under CARMA as a dynamic population game (DPG), in which a stationary Nash equilibrium (SNE) is guaranteed to exist. Unlike existing monetary schemes, CARMA is demonstrated, both analytically and numerically, to achieve (a) an equitable traffic assignment with respect to heterogeneous income classes and (b) a strong Pareto improvement in the long-term average travel disutility with respect to no policy intervention. With extensive numerical analysis, we show that CARMA is able to retain the same congestion reduction as an optimal monetary tolling scheme under uniform karma redistribution and even outperform tolling under a well-designed redistribution scheme. We also highlight the privacy-preserving feature of CARMA, that is, its ability to tailor to the private preferences of commuters without centrally collecting the information. History: This paper has been accepted for the Transportation Science Special Issue on TSL Conference 2023. Funding: This work was supported by NCCR Automation, a National Centre of Competence in Research, funded by the Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung [Grant 180545]. Supplemental Material: The online appendices are available at https://doi.org/10.1287/trsc.2023.0323 .

In many real-world large-scale decision problems, self-interested agents have individual dynamics and optimize their own long-term payoffs. Important examples include the competitive access to shared resources (e.g., roads, energy, or bandwidth) but also non-engineering domains like epidemic propagation and control. These problems are natural to model as mean-field games. Existing mathematical formulations of mean field games have had limited applicability in practice, since they require solving non-standard initial-terminal-value problems that are tractable only in limited special cases. In this letter, we propose a novel formulation, along with computational tools, for a practically relevant class of Dynamic Population Games (DPGs), which correspond to discrete-time, finite-state-and-action, stationary mean-field games. Our main contribution is a mathematical reduction of Stationary Nash Equilibria (SNE) in DPGs to standard Nash Equilibria (NE) in static population games. This reduction is leveraged to guarantee the existence of a SNE, develop an evolutionary dynamics-based SNE computation algorithm, and derive simple conditions that guarantee stability and uniqueness of the SNE. We provide two examples of applications: fair resource allocation with heterogeneous agents and control of epidemic propagation. Open source software for SNE computation: https://gitlab.ethz.ch/elokdae/dynamic-population-games .