AVEC ’14 Tire Force Estimation Utilizing Wheel Torque Measurements and Validation in Simulations and Experiments *Anton Albinsson (Corresponding author), *Fredrik Bruzelius, **Mats Jonasson,*Bengt Jacobson *Chalmers Technical University 412 96 Gothenburg Sweden Phone: +46 (0)31 772 10 00 E-mail: anton.albinsson@chalmers.se, fredrik.bruzelius@chalmers.se, bengt.jacobson@chalmers.se **Volvo Cars 405 31Gothenburg Sweden Phone: +46 (0)31 59 00 00 E-mail: mats.jonasson@volvocars.com This study investigates a new tire force estimator based on the recursive least square (RLS) method. Tire force estimation with known driving wheel torque is studied and compared to the case with torque estimation from the internal combustion engine. This is motivated by a future scenario with electric propulsion, which reasonably gives improved wheel torque estimations. Sensitivity to vehicle parameters and challenges with individual lateral tire force estimation are also investigated. The results, experimental and simulation data, show good performance and potential for tire force estimation using the RLS method. Topics / Modeling and simulations, State estimation 1. INTRODUCTION Active safety systems are becoming more advanced in order to meet higher requirements from both the market and legislations. With more accurate and new information about the vehicle states, the current and future active safety systems can be improved. The tire forces are important vehicle states that can be used for vehicle parameter estimation and tire identification online. For electric vehicles new opportunities exists for estimation and for improving active safety systems. The main advantage with electric motors, in an estimation context, is that the delivered torque to the wheels is more straightforward to estimate compared to the torque from an Internal Combustion Engine (ICE). Additionally, electric propulsion is typically also used for braking and can be wheel individual, compared to traditional propulsion. Electric motors can then be considered as extra sensors that can be utilized. This study investigates how a better estimation of the applied wheel torque can improve both lateral and longitudinal tire force estimation. In [1,2] both lateral and longitudinal tire forces are estimated but no reliable estimate of the wheel torque is available and hence no individual longitudinal tire force estimation is attempted. [3,4] utilizes the wheel dynamics and the driving wheel torque to estimate the longitudinal tire forces but the validation is done using simulations only. The present study highlights problems associated with tire force estimation, such as uncertain parameters and individual tire force estimation. Individual lateral tire force estimation has been discussed in [1,2,4]. These estimations works well under special conditions constrained by the assumptions made, e.g. uniform friction, low level of lateral acceleration. [5] use a large number of vehicle parameters which is not possible to estimate online and the estimator is prone to be very vehicle specific. A more thorough discussion about the parameters affecting the individual tire force estimation is presented here. The estimator is based on a relatively simple Recursive Least Square (RLS) method compared to the Kalman filter methods used in [1-5]. One of the main reasons for estimating tire forces is to combine the data with a slip estimator in order to identify vehicle and tire parameters. The tire force estimates are to be considered as inputs to the tire model. Including a tire model in the tire force estimator would thus not add any new information to the system unless the tires are identified offline. No tire model has been used to estimate the tire forces in this study which makes the estimator robust to variations in tire parameters and road surfaces. Validation of the estimator was done with simulations using IPG Carmaker and data from real experiments. 2. ESTIMATOR STRUCTURE The input to the estimator is the lateral and longitudinal body acceleration, yaw rate, wheel speeds on all four wheels, steering wheel angle and the propulsion torque on the front wheels. The test data used in the study is from tests performed using a front wheel drive vehicle equipped with an internal combustion engine and not an electrical motor. The test vehicle was fitted with torque wheels for all tests. As a result of the vehicle not being equipped with electric motors, the torque measurements from the torque 12th International Symposium on Advanced Vehicle Control September 22-26, 2014 294 Copyright © 2014 Society of Automotive Engineers of Japan, Inc. All rights reserved 20149250