Nonlinear Model Predictive Control for Drifting on Low-Friction Surfaces

Rin Yonetani･Han Wen･Hiroyuki Okuda･Takuma Yamaguchi･Tatsuya Suzuki (Nagoya University)

When a vehicle enters a low-friction road surface such as a snowy or gravel road, the tires may skid and the vehicle may become unstable. Therefore, it is thought that safety can be ensured if it is possible to freely control drifting, which is cornering by intentionally sliding the rear wheels. In this study, we propose a control system for driving in a drift state using nonlinear model predictive control.

Lane-Changing Space Selection at Congested Merging Area

Keiju Nishimura･Hanwool Woo (Kogakuin University)

This study assumes a scene in which the ego vehicle performs a lane change into the congested main lane. The proposed method selects the lane-changing space based on the remained distance of the merging lane and relative amounts of adjacent vehicles on the main lane. In addition, our method estimates the acceptability of the vehicles on the main lane with respect to the lane-changing of the autonomous vehicle. This enables a more flexible decision making for various driver characteristics of the main lane drivers. Through simulations, the effectiveness of the proposed method is evaluated.

Prediction of 4G LTE Throughput in Vehicles using LSTM

Xiangqing Zhang･Hidenori Yamashita･Michikazu Umemura (AutoNetworks Technologies)

In this presentation, we will introduce a technology for predicting future 4G communication speeds in vehicles using LSTM. We have achieved high-precision predictions by utilizing past communication quality records and geographical location information. Based on these predictions, we can avoid communication issues and thereby contribute to improving the overall stability of communication services in vehicles. This technology holds great potential for applications in fields that require fast and stable communication, such as remote driving.

A Method of Controlling a Dual-Structure Electric Steering System for Autonomous Driving

Taehong Kim (Hyundai Mobis)

As autonomous driving technology comes to reality, the role of steering, one of the key systems of the vehicle, is becoming more important. In particular, the position control performance of the electric steering system plays a key role in terms of safety and convenience.
In particular, Lv.4 autonomous driving is not mandatory for the driver to look forward and the driver does not ride, so it is essential to meet ASIL-D by default. For this, a fully redundant system is required. Fully redundant EPS basically has all components duplicated based on the power pack standard. That is, electronic devices, ECUs, sensors, and motor winding are all duplicated. Based on this structure, an optimized algorithm is needed to successfully perform position control by synchronizing position control by receiving a redundant command steering angle from the autonomous driving module.