Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


Anisha_thesis_final document.pdf (11.6 MB)

ETD Abstract Container

Abstract Header

Safety by Design in Adaptive Cruise Control using Hamilton Jacobi Reachability Analysis

Karthyedath, Anisha
http://rave.ohiolink.edu/etdc/view?acc_num=osu1669892816441656

Abstract Details

2022, Master of Science, Ohio State University, Electrical and Computer Engineering.
We are in the process of successfully integrating various autonomous systems into our day- to-day life. However, ensuring the safe performance of these systems is challenging because the deployment of a complex system with a lot of environmental uncertainties while guaranteeing safety always is impossible. In addition, the advancement in connectivity and sensor technology making different safety and security problems in the safety critical systems. The focus of this thesis is on designing a safety-by-design controller using Hamilton Jacobi Reachability (HJR) Analysis which guarantees the safety properties of the system without further verification. The Hamilton Jacobi Reachability analysis is applied to compute the reachable set and the value function which will be then used for the design of the control laws that satisfy the defined safety specification of the system. The Adaptive Cruise Control (ACC) system which is one of the safety critical features of the Advanced Driver Assistance System (ADAS) is used as a motivating application for this thesis. The ACC system focuses mainly on driver comfort and safety. However, most of the ACC system lacks fully developed collision avoidance features which are essential when considering the safety of the system in the first place. The sudden deceleration of the lead vehicle, slower vehicle cut-in scenario and other adversarial attacks are some examples where safety should come first. The goal here is to design an ACC system using HJR analysis which will enhance the performance of the standard proportional integral controlled ACC system. The designed ACC system provides more control authority to the vehicle when the driver is not able to react fast enough to sudden collision scenarios. The safe controller is designed on top of the standard proportional- integral controller. That is the safety by design controller consisting of PI controller as performance controller and HJR controller as safety controller. The safety controller will act only on the ACC system when the system is about to violate the safety specification. This thesis formulated the collision problem between the ego vehicle which is equipped with the ACC system and the lead vehicle as a pursuit-evasion game by considering the worst-case scenario. Fully braking of the leading vehicle is considered the worst-case scenario while the ego vehicle tries to avoid it. The unsafe set and the value function are computed by solving this game. Based on the value function control signal is computed during the collision for the safe controller. The ego vehicle can always avoid a collision if it is outside of this unsafe set of states. The designed controller is verified based on a different scenario which includes some external attacks that possibly happened to ACC-equipped vehicles. Moreover, the performance of the designed controller is compared with the standard proportional-integral (PI) controlled ACC system, and the Model Predictive Control based ACC system. In the normal operation scenario case, the mean squared error in the distance of the ACC system with PI is 31.84 and is reduced to 24.532 using PI with HJR and which makes the PI with HJR controller better in terms of performance. Similarly, in the vehicle cut in front of the ACC-equipped vehicle case, the MSE in distance is decreased from 111.11 to 48.26 when using PI with HJR. In the case of a denial-of-service attack also called a random signal-dropping attack, the ACC system designed using PI and HJR can withstand a 15% control signal drop. However, the PI-based controller can only withstand 9.7% of the signal drop. The designed safety by design controller enhanced the performance of the standard PI controller and exhibit similar performance to that of the model predictive controller.
Abhishek Gupta (Committee Member)
Qadeer Ahmed (Advisor)
119 p.
Electrical Engineering
Adaptive cruise control, Hamilton Jacobi Rechability

Recommended Citations

Citations

  • Karthyedath, A. (2022). Safety by Design in Adaptive Cruise Control using Hamilton Jacobi Reachability Analysis [Master's thesis, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1669892816441656

    APA Style (7th edition)

  • Karthyedath, Anisha. Safety by Design in Adaptive Cruise Control using Hamilton Jacobi Reachability Analysis. 2022. Ohio State University, Master's thesis. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1669892816441656.

    MLA Style (8th edition)

  • Karthyedath, Anisha. "Safety by Design in Adaptive Cruise Control using Hamilton Jacobi Reachability Analysis." Master's thesis, Ohio State University, 2022. http://rave.ohiolink.edu/etdc/view?acc_num=osu1669892816441656

    Chicago Manual of Style (17th edition)

osu1669892816441656
100
© 2022, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.