PostDoc Project
Title:
Development and Validation of Intelligent Estimation Algorithms for Connected and Autonomous Vehicles
Dates:
2025/03/25 - 2025/08/31
Student:
Supervisor(s):
Description:
The project ArtISMo aims to develop intelligent estimation algorithms to enhance the control and safety of connected and autonomous vehicles (CAVs). Real-time knowledge of various variables such as distances, speeds, and angles of movement is essential to ensure reliable and safe autonomous driving. Indeed, autonomous vehicles must face significant uncertainties, including sensor defects, variations in system parameters (such as tire characteristics), and potential attacks. Non-linear observers will enable the reconstruction of state variables in real-time, providing critical information for control and diagnostics. This will contribute to:
⬢ Improving safety by estimating unmeasurable internal states, thus allowing for rapid reactions to anomalies.
⬢ Optimizing performance by dynamically adjusting models based on real driving conditions and data collected in real-time.
⬢ Facilitating innovation by exploring novel learning methods to model the complex components of the vehicle better.
Objectives:
The primary objective is to develop intelligent estimation algorithms that integrate both classical theoretical automatic control methods and online learning techniques to address the complexities of vehicle dynamic behavior and accurately estimate the state variables necessary for controlling CAVs [1],[2],[3].
In the second phase, we will implement the developed observers based on real data to validate their performance and robustness under various operating conditions. The goal of using data collected from multiple vehicles is to identify trends and behaviors that may not be apparent from a single vehicle's data. This makes it possible to refine the models and improve the accuracy of estimates by leveraging a wide range of data. The algorithms will first be tested on a 3D simulator (CARLA) for initial validation, and then on the existing LIMO mobile robot at CRAN (see figure below).
A key goal of this position is to publish the research findings in journals and at scientific events in the field.
References:
[1] Jeon, W., Chakrabarty, A., Zemouche, A., & Rajamani, R. (2021). Simultaneous state estimation and tire model learning for autonomous vehicle applications. IEEE/ASME Transactions on Mechatronics, 26(4), 1941-1950.
[2] Fu, J., Wen, G., Xu, Y., Zemouche, A., & Zhang, F. (2022). Resilient cooperative control of input-constrained networked cyber-physical systems. In Security and Resilience in Cyber-Physical Systems: Detection, Estimation, and Control (pp. 267-298). Cham: Springer International Publishing.
[3] Bessafa, H., Delattre, C., Belkhatir, Z., Zemouche, A., & Rajamani, R. (2024, July). Radar sensor-based longitudinal motion estimation by using a generalized high-gain observer. In American control conference, ACC 2024.
⬢ Improving safety by estimating unmeasurable internal states, thus allowing for rapid reactions to anomalies.
⬢ Optimizing performance by dynamically adjusting models based on real driving conditions and data collected in real-time.
⬢ Facilitating innovation by exploring novel learning methods to model the complex components of the vehicle better.
Objectives:
The primary objective is to develop intelligent estimation algorithms that integrate both classical theoretical automatic control methods and online learning techniques to address the complexities of vehicle dynamic behavior and accurately estimate the state variables necessary for controlling CAVs [1],[2],[3].
In the second phase, we will implement the developed observers based on real data to validate their performance and robustness under various operating conditions. The goal of using data collected from multiple vehicles is to identify trends and behaviors that may not be apparent from a single vehicle's data. This makes it possible to refine the models and improve the accuracy of estimates by leveraging a wide range of data. The algorithms will first be tested on a 3D simulator (CARLA) for initial validation, and then on the existing LIMO mobile robot at CRAN (see figure below).
A key goal of this position is to publish the research findings in journals and at scientific events in the field.
References:
[1] Jeon, W., Chakrabarty, A., Zemouche, A., & Rajamani, R. (2021). Simultaneous state estimation and tire model learning for autonomous vehicle applications. IEEE/ASME Transactions on Mechatronics, 26(4), 1941-1950.
[2] Fu, J., Wen, G., Xu, Y., Zemouche, A., & Zhang, F. (2022). Resilient cooperative control of input-constrained networked cyber-physical systems. In Security and Resilience in Cyber-Physical Systems: Detection, Estimation, and Control (pp. 267-298). Cham: Springer International Publishing.
[3] Bessafa, H., Delattre, C., Belkhatir, Z., Zemouche, A., & Rajamani, R. (2024, July). Radar sensor-based longitudinal motion estimation by using a generalized high-gain observer. In American control conference, ACC 2024.
Keywords:
Nonlinear estimation, Vehicle dynamics, Learning
Department(s):
| Control Identification Diagnosis |