Assessment Brief Automotive Systems Engineering Msc Module: 19TTP409 Assessment Title: CW2: Kalman Filter and Adaptive Cruise Control Module Leader: Cunjia Liu Responsible Academic: Cunjia Liu Hand out date: 07 Feb 2020 Due date: 29 April 2020 Return of preliminary marks: 21 May 2020 Credit weight: 50% Coursework elements, submission & marking process 1. A report of 6 pages, excluding title, content & appendix (11pt text & 3cm margins give about 500-600 words per page) To be submitted electronically on learn.lboro.ac.uk In PDF format, A4, can use colour Marked manually according to the marking scheme General Guidelines Please see our coursework code of conduct at https://learn.lboro.ac.uk/course/view.php?id=3930§ion=5 for advice on formatting, deadline extensions, plagiarism rules, and report writing advice. By submitting your coursework, you acknowledge the coursework code of conduct. Marking Guidelines Only the report will be marked. The Simulink code needs to be submitted to the Learn system in case we need to check the integrity of your code. The mark allocation is given as follows: 1. Algorithm design: a. Kalman filter design: 30%. b. Control switching logic design: 20%. c. The integration of Kalman filter and adaptive cruise control: 10%. 2. Discussion on algorithm performance: 30% (each above point gets 10%). 3. Report structure and presentation: 10%. A marking guidance is also available on the module Learn page Support Generic feedback for the previous cohort is available on https://learn.lboro.ac.uk For questions, please use the discussion forum on learn, or email Anoma Malalasekera
[email protected]. 19TTP409 Coursework 2: Kalman Filter & Adaptive Cruise Control Introduction Consider the Adaptive Cruise Control scenario where the ego vehicle is trying to maintain a prescribed speed while maintaining a safe distance from the vehicle in the front (denoted as lead vehicle). This coursework requires 1) to design and implement a Kalman filter algorithm to estimate the lead vehicle’s position, velocity and acceleration using the noise corrupted sensor data of the lead vehicle; 2) to complete the control strategy so that it can switch between normal cruise control mode and space control mode and take the estimated signals for control. The simulation environment is provided in the following Simulink model along with a m-file, which can be downloaded from Learn. The basic function of this Simulink model and principle of adaptive cruise control will be covered in the lecture. Instead of directly using the actual lead vehicle position and velocity for feedback control design, this coursework assumes that an onboard sensor (e.g. Laser range finder) will be used to measure the lead vehicle’s position. The sensor reading is corrupted by a Gaussian noise with zero mean and variance of 1 m2,, i.e. = + , where is the lead vehicle position and is the noise. Therefore, to make sure the original control strategy is still functioning, you need to design a Kalman filter to estimate at least the vehicle’s position and velocity from noise corrupted measurements. The ego vehicle’s states are assumed to be known accurately. The speed profile of the lead vehicle is included in the Simulink model. All other parameters have been specified in the associated m-file and Simulink file. Note 1: one dimensional 3rd order vehicle model, known as the constant acceleration model, can be used in Kalman filter design, so that it can capture the velocity and acceleration changes of the lead vehicle. For example: +1 = + + 22 + 2 2 , +1 = + + and +1 = + , where , and are vehicle position, velocity and acceleration, respectively, is the sampling interval and is the process noise, which also follows a Gaussian distribution of zero mean. The variance of the process noise can be tuned based on the performance. Note 2: you can set the initial state of the Kalman filter at a reasonable value (e.g. [45 20 0]) and explain your choice of the initial covariance matrix; Note 3: The basic design principle of the switching logic for the controller can be found in the lecture note. For the coursework, the desired conditions for spacing control are: • IF {actual distance < desired distance}, THEN {do spacing control} • IF { (desired distance < actual distance < desired distance + a buff distance) AND (lead vehicle speed < ego vehicle set speed) }, THEN {do spacing control} Requirement The task of this coursework is first to design a Kalman filter to provide the estimates of the lead vehicle’s states that can be used in the controller to achieve adaptive cruise control. Then, the control strategy needs to be completed by designing the switch logic between the cruise control mode and space control mode. A written report no more than 6 pages is required from this coursework. The report should 1) describe the design of algorithms in details, in a way that other people can use it as an instruction to recreate your algorithm (you can use diagram if necessary); 2) demonstrate the performance of your Kalman filter algorithm by comparing the estimated lead vehicle states against the actual state; 3) demonstrate the performance of your adaptive cruise control performance in conjunction with the Kalman filter; 4) discuss the advantages and disadvantages of your algorithms and how you tune those critical parameters in your algorithm to achieve a good performance; 5) discuss any other way that the adaptive cruise control strategy can be improved. Only the CW report will be marked. The Simulink model needs to be submitted to the system in case we need to check the integrity of your programming. The mark allocation is given as follows: 1. Algorithm design: a. Kalman filter design: 30%. b. Control switching logic: 20%. c. The integration of Kalman filter and adaptive cruise control: 10%. 2. Results discussion: 30%. 3. Report structure and presentation: 10%. Submission Please submit the report together with your code to Learn before the deadline. Make sure your Matlab/Simulink can be run by just click on button (otherwise write a ‘Readme’ file). Your codes will be used as references but will only be checked if necessary. Marking criteria 1. Algorithm: 60% A. (excellent) The algorithms are working well, and the descriptions of the algorithms are rigorous and mathematically sound. B. (good) The algorithms are working correctly, and the other people can use the description as an instruction to recreate the algorithm. C. (satisfactory) The algorithms are working correctly, the descriptions of the algorithms are clear, but some details may be missing. D. (unsatisfactory) Very basic or incomplete description of the algorithm. Scrappy / incoherent text. Report padded out with irrelevant detail. Figures poorly presented / incomplete / vague. 2. Results and discussion: 30% A. (excellent) Clear demonstration of your simulation data. Be able to identify and discuss the advantages and disadvantages of your algorithms and find the effective way of tuning the parameters. Clear and fair conclusions have been drawn with promising suggestions for future improvement. B. (good) Clear demonstration of your simulation data. Discussed the advantages and disadvantages of your algorithm and showed the influence of critical parameters on estimation performance. Fair conclusions have been drawn. C. (satisfactory) Be able to demonstrate your simulation data, but some technical aspects are missing. Discussed the advantages and disadvantages of your algorithm but may not have sufficient discussions on the influence of parameters. Fair conclusions have been drawn. D. (unsatisfactory) The results are not presented in a logic way and the read cannot draw the conclusion. Did not manage to analyse the performance of this algorithm and/or point out the features of the algorithm. The conclusions may not be meaningful or clear. 3. Structure and expression: 10% A. (excellent) Arguments are clear within each well-ordered section. Clear presentation of all data using figures and appropriate use of Appendices if necessary. Concise text with some impression that not all results could be included. B. (good) Good presentation of analysis. Sensibly ordered and complete sections. Figures and data all fully complete. Easily readable text. C. (satisfactory) Essential components of analysis given. Comprehensible text. Some disjointing of descriptions. Figures may not have all detail provided. D. (unsatisfactory) Very basic or incomplete analysis. Scrappy / incoherent text. Report padded out with irrelevant detail. Figures poorly presented / incomplete / vague.
