Scott Moura’s research while affiliated with University of California, Berkeley and other places

What is this page?

This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (56)

Fig. 1: Vehicle-in-the-loop control architecture [12]
Fig. 2: Lane selection problem in urban road
Fig. 3: Comparison of quadratic energy consumption model and measured data
Fig. 4: Illustration of a graph for navigation through multiple traffic lights. The graph approximates the pass or nonpass decision for a fixed lane and velocity
Fig. 5: The virtual CARLA simulator map, the satellite image of the actual testing site, and the physical test vehicle

+1

Energy-Aware Lane Planning for Connected Electric Vehicles in Urban Traffic: Design and Vehicle-in-the-Loop Validation
  • Preprint
  • File available

March 2025

·

86 Reads

·

·

Eunhyek Joa

·

[...]

·

Francesco Borrelli

Urban driving with connected and automated vehicles (CAVs) offers potential for energy savings, yet most eco-driving strategies focus solely on longitudinal speed control within a single lane. This neglects the significant impact of lateral decisions, such as lane changes, on overall energy efficiency-especially in environments with traffic signals and heterogeneous traffic flow. To address this gap, we propose a novel energy-aware motion planning framework that jointly optimizes longitudinal speed and lateral lane-change decisions using vehicle-to-infrastructure (V2I) communication. Our approach estimates long-term energy costs using a graph-based approximation and solves short-horizon optimal control problems under traffic constraints. Using a data-driven energy model calibrated to an actual battery electric vehicle, we demonstrate with vehicle-in-the-loop experiments that our method reduces motion energy consumption by up to 24% compared to a human driver, highlighting the potential of connectivity-enabled planning for sustainable urban autonomy.

Download
Fig. 2 Data generation and analysis. (a) Aging trajectories under 4 verification scenarios, i.e., 25, 35, 45 and 55 1C. (b) Illustration of the multi-step charging profile. The cut-off voltage values, i.e., from U1 to U9. (c) Stepwise state of charge (SOC) increments and their accumulation. (d) Visualization of cut-off voltage. (e) Temperature-wise trends of cut-off voltage values each charging step averaged over all batteries. (f) The deviation of cut-off voltage values at each temperature, defined as initial manufacturing variabilities (IMVs), for each charging step. IMVs are calculated over batteries at a given temperature. (g) Battery lifetime distribution under EOL73, with lifetime deviations indicated. EOL73 is defined as 73% of the nominal capacity. (h) Arrhenius plot of 4 temperature verification scenarios. (i) Arrhenius transferability (AT) score quantifies dissimilarities between accessible data and early data of batteries to be verified. r s is the source domain aging rate, r t is the target domain aging rate, E a is the activation energy, k B is the Boltzmann constant, T s is the source domain temperature and T t is the target domain temperature.
Fig. 3 Featurization taxonomy. (a) Charging dynamics in different verification scenarios, i.e., 25, 35, 45, 55 1C. (b) The fundamental electrochemical principle for featurization taxonomy. (c) Featurization taxonomy, including initial manufacturing variabilities (IMVs) and multi-dimensional chemical processes. (d) Visualization of features and battery-wise variations in the lifetime direction. The color maps normalized feature values, and the size of bubbles maps the deviations across battery instances.
Fig. 4 The prototype verification results. Parity plot of lifetime prediction under (a) 35 and (b) 45 1C verification scenarios (intermediate temperature verification), with battery-wise prediction deviation presented. All numerical results achieved in this plot are under the early verification setting that uses 20% of early data from batteries to be verified, otherwise specified. Experimental settings for (a)-(e) are multi-source domain adaptation, i.e., data at 25 and 55 1C is accessible. (c) Model performance comparison between this work and long-short-term memory (LSTM) neural network, model without considering initial manufacturing variabilities (IMVs), model without physics-informed machine learning, and empirical formula. (d) Model sensitivity under different end-of-life capacities. Model prediction error against (e) early data access and (f) parallel guiding sample requirements of target verification scenarios. Accessible data at 55 1C is assumed, which is used to predict the lifetime of 25, 35, and 45 1C verification scenarios. (g) Feature importance of capacity features (Q) in different charging stages. (h) Feature importance of non-capacity features by charging stages, listed in a stepwise order.
Fig. 5 Decoupling battery degradation loss and polarization types. (a) Schematic illustration of battery performance degradation in different lifetime stages. (b) Battery degradation modes and their loss type, i.e., thermodynamics and kinetics. (c) Li-ion concentration visualization inside the battery, i.e., from the anode (Ano.), separator (Sep.) to the cathode (Cat.) under multi-step charging. (d) Overpotential evolution in a lifetime direction from the initial cycle to the 1000th cycle under multi-step charging. (e) Degradation dominance evolution at 25, 35, 45, and 55 1C, respectively, which is calculated by the cycle-wise SAGE importance of features describing relevant microscopic degradation behaviors. (f) Temporally resolved correlation between the thermodynamic loss and concentration polarization. (g) Incremental capacity analysis of discharging curves. (h) Normalized capacity loss types, i.e., thermodynamic and kinetic types, for the unit state of charge (SOC) at 25, 35, 45, and 55 1C, respectively. Proportion comparison of thermodynamic (85%) and kinetic (15%) loss types, averaged over all temperatures. The machine learning insight, i.e., the contribution of thermodynamic loss (79%) is indicated. (i) Normalized polarization types, i.e., concentration and other (ohmic and electrochemical) polarization, at 25, 35, 45, and 55 1C, respectively. The proportion of concentration (82%) and other (18%) polarization averaged over all temperatures. The machine learning insight, i.e., the contribution of concentration polarization (74%) is indicated.
Non-destructive degradation pattern decoupling for early battery trajectory prediction via physics-informed learning

January 2025

·

633 Reads

·

2 Citations

Energy & Environmental Science

·

·

·

[...]

·

Guangmin Zhou

The paper proposes a physics-informed model to predict battery lifetime trajectories by computing thermodynamic and kinetic parameters, saving costly data that has not been established for sustainable manufacturing, reuse, and recycling.

Download
Remaining Discharge Energy Prediction for Lithium-Ion Batteries Over Broad Current Ranges: A Machine Learning Approach

April 2024

·

76 Reads

·

·

Scott Moura

·

[...]

·

Huazhen Fang

Lithium-ion batteries have found their way into myriad sectors of industry to drive electrification, decarbonization, and sustainability. A crucial aspect in ensuring their safe and optimal performance is monitoring their energy state. In this paper, we present the first study on predicting the remaining energy of a battery cell undergoing discharge over wide current ranges from low to high Crates. The complexity of the challenge arises from the cell's Crate dependent energy availability as well as its intricate electro-thermal dynamics. To address this, we introduce a new definition of remaining discharge energy and then undertake a systematic effort in harnessing the power of machine learning to enable its prediction. Our effort includes two parts in cascade. First, we develop an accurate dynamic model based on integration of physics with machine learning to capture a battery's voltage and temperature behaviors. Second, based on the model, we propose a machine learning approach to predict the remaining discharge energy under arbitrary Crates and pre-specified cutoff limits in voltage and temperature. The results from our experiments show that the proposed approach offers high prediction accuracy and amenability to training and computation.

Download
A nonlinear fractional-order dynamical framework for state of charge estimation of LiFePO4 batteries in electric vehicles

October 2023

·

182 Reads

·

Scott Moura

·

Dylan Kato

·

An efficient state of charge (SOC) estimation for LiFePO4 batteries in electric vehicles (EVs) has been an open problem so far, largely due to its non-measurable nature. This paper tackles this problem by presenting a fractional-order (FO) dynamical framework to unravel and understand the inherent dynamics of the LiFePO4 battery which leads to an improved estimation of SOC. First, a FO model (FOM) is proposed where the parameters are introduced as nonlinear functionalities of SOC. It has been observed that the FO defined as a nonlinear function of SOC is crucial in identifying its progression during the weakly measurable flat, open circuit curve of the battery; a property the integer order models (IOMs) fail to capture. Second, a fractional order estimator (FOE) is designed incorporating the SOC based nonlinearities of the model parameters. The FO derivative being a memory-based operator improves estimation as it can store historical information of the speed profiles of the EV. The proposed framework of nonlinear FOM and FOE design is validated through both simulation and experimental results. Precise estimation of the battery parameters using the proposed framework can be applied to protect the battery management system, mitigate overcharge or discharge, prevent hazardous accidents, and enhance battery life, eventually leading to an energy-efficient mode of green transportation.

Download

Citations (29)

... Thermodynamic and dynamic models of battery degradation were incorporated into a physics-informed machine learning framework, enabling non-destructive decoupling of degradation modes and lifecycle prediction. Compared to traditional methods, the model achieved a 25-fold increase in prediction speed and 95.1% accuracy, significantly reducing validation costs [165]. ...

Reference:

A Survey of Physics-Informed AI for Complex Urban Systems
Non-destructive degradation pattern decoupling for early battery trajectory prediction via physics-informed learning

Energy & Environmental Science

·

·

·

[...]

·

Guangmin Zhou

... This is often referred to as coupling between the power and transport systems, where we extend vehicle routing problems to include charging and energy constraints. Approaches also vary on their objective; charging as fast as possible [20], only at their home base [21], minimizing energy consumption [22], or minimizing charging costs [16]. In general, EV charging is considered only to be able to provide demand flexibility over shorter durations (hours) [23], [24]. ...

Reference:

On the Potential of Electrified Supply Chains to Provide Long Duration Demand Flexibility
Saving energy with eco-friendly routing of an electric vehicle fleet
  • Citing Article

  • September 2024

Transportation Research Part E Logistics and Transportation Review

Soomin Woo

·

·

Scott J. Moura

·

Francesco Borrelli

... Zhang et al. (2022) presented a coordinated restoration optimization method for distribution networks that integrates network reconfiguration, emergency power vehicles (EPVs), and uncertain E-taxi capabilities, utilizing a bilevel programming model to enhance load restoration effectiveness. Zeng et al. (2023) studied the integration of energy services from shared plug-in electric vehicles into a mobility-on-demand platform. Sayarshad (2025b) assessed the capacity of aggregator fleets to improve the reliability of the electricity grid in the face of cyber threats. ...

Reference:

Securing Power Grids and Charging Infrastructure: Cyberattack Resilience and Vehicle-to-Grid Integration
Joint Mobility and Vehicle-to-Grid Coordination in Rebalancing Shared Mobility-on-Demand Systems
  • Citing Article

  • January 2023

IFAC-PapersOnLine

·

Scott Moura

·

... Reference[57] demonstrated how ESS and RES reduce their impact on the electrical grid. Reference[58] addressed the nonlinear elements in the battery degradation model and charging pattern; two linearization strategies are recommended for use with MILP. With particular attention to the CS variables, design, techniques for operation, and management functions, Ref[59] highlighted a photovoltaic dc fast charging station (PVDCFCS). ...

Reference:

Decentralized Energy Solutions: The Impact of Smart Grid-Enabled EV Charging Stations
Electric fleet charging management considering battery degradation and nonlinear charging profile

Energy

·

·

Scott Moura

... In [65], neural networks are trained to classify thermal safety/risk levels. Integrating physics with neural networks can further enhance the fault detection accuracy [66,67], with the idea tracing to [68]. Riding on the advances in machine learning, other techniques, such as transfer learning, Gaussian process regression and support vector machines, have found their way into thermal fault prediction or detection [69][70][71]. ...

Reference:

BattBee: Equivalent Circuit Modeling and Early Detection of Thermal Runaway Triggered by Internal Short Circuits for Lithium-Ion Batteries
Integrating physics-based modeling with machine learning for lithium-ion batteries
  • Citing Article

  • January 2023

Applied Energy

·

Scott Moura

·

·

Huazhen Fang

... For example, in a Tesla Model S, there are 96 series-connected battery modules, so the workload and computational cost can be very high if we adopt a complex filter-based method for each module [27]. To solve such a problem, we noticed that some studies proposed to use a simple algorithm first to identify the "weakest" cell in the battery pack (i.e., the cell that has the lowest voltage or some other apparent characteristics that make it more likely to have the lowest SOC or SOH) and then use another complex algorithm to estimate the state of those weakest cells [27,28,29,30]. While this idea can partially solve the question, the algorithm may not capture the weakest cell correctly, which could make the estimation too optimistic. ...

Reference:

Relax, Estimate, and Track: a Simple Battery State-of-charge and State-of-health Estimation Method
State-Of-Health Estimation Pipeline for Li-ion Battery Packs with Heterogeneous Cells
  • Citing Conference Paper

  • June 2022

·

·

·

[...]

·

Scott Moura

... Smiley and Plett (2018) presented NN adaptive physics-based reduced-order model of an aged lithium-ion cell, selected using an interacting multiple-model Kalman filter. Pozzi et al. presented a neural network-based approximation of model predictive control for a lithium-ion battery with electro-thermal dynamics (Pozzi et al. 2022). ...

Reference:

A robust non-linear method for the state-of-health estimation for lithium-ion batteries based on dissipativity theory for electric vehicle applications
A Neural Network-Based Approximation of Model Predictive Control for a Lithium-Ion Battery with Electro-Thermal Dynamics

·

Scott Moura

·

... Battery manufacturing management Interdependent parameters, processes, and stages Meiners et al. 2022) Understanding the different production stages ) Mathematical modeling (Tu et al. 2023) Determination of the product quality (Stock et al. 2022) Battery supply chain (Egbue and Long 2012; Determination of quality parameters (Schnell et al. 2019) Sustainable battery supply Innovation and optimization (Mao et al. 2021) Resource allocation (Miao et al. 2022) Environmental protection (Miao et al. 2022 Content courtesy of Springer Nature, terms of use apply. Rights reserved. ...

Reference:

Sustainable innovation in the lithium industry through data science: perspectives and challenges
Integrating Physics-Based Modeling with Machine Learning for Lithium-Ion Batteries
  • Citing Article

  • January 2022

SSRN Electronic Journal

·

Scott Moura

·

·

... It is further influenced by a number of uncertainty factors such as road conditions, measurement accuracy, and a long tail of behavioral uncertainty of on-road agents. However, if executed efficiently, lane changing coupled with speed adjustment can yield significant improvement in minimizing overall travel time while ensuring passenger comfort [2]. ...

Reference:

SLAS: Speed and Lane Advisory System for Highway Navigation
Risk-Aware Lane Selection on Highway with Dynamic Obstacles
  • Citing Conference Paper

  • July 2021

·

David Isele

·

Kikuo Fujimura

·

Scott J. Moura

... The first electrochemical hybrid residual model was developed in simulation in [11], in which a recurrent neural network was implemented to predict the residual between a full-order DFN model and reduced-order SPM. In [12], state information from the electrochemical model was incorporated as an input to the ML residual model, which was found to significantly improve the residual prediction accuracy in simulation. This work was continued in [13] with a theoretical justification for the use of state information as an ML input. ...

Reference:

Joint Parameterization of Hybrid Physics-Based and Machine Learning Li-Ion Battery Model
Integrating Electrochemical Modeling with Machine Learning for Lithium-Ion Batteries
  • Citing Conference Paper

  • May 2021

·

Scott Moura

·