International Journal of Vehicle Autonomous Systems (8 papers in press)
The safety potential of automatic emergency braking and adaptive cruise control and actions to improve the potential
by Roni Utriainen, Markus Utriainen And Pöllänen
Abstract: The study investigates the potential of automatic emergency braking (AEB)
and adaptive cruise control (ACC) systems to prevent fatal rear-end, intersection and
pedestrian crashes in Finland. The systems' possibilities to prevent crashes were
assessed using data on 115 in-depth investigated fatal crashes. The data includes all
fatal crashes in the three studied crash types in 2014-2016. This study considers the
impact of estimated speed, weather conditions and intentionality on the systems
operation. AEB and ACC could potentially have prevented 41% of the crashes. The
highest safety potential in terms of share of hypothetically prevented crashes was
recognised in rear-end (45%) and pedestrian crashes (45%) and the lowest in
intersection crashes (36%). This study complements previous research, which amount is
low especially considering the potential to reduce pedestrian and intersection crashes, and which has typically been limited in the aspects that are considered in analysing the safety potential. Additionally, issues related to systems operational conditions are discussed and the possibilities to further increase the safety potential are assessed.
Keywords: automatic emergency braking; AEB; adaptive cruise control; ACC; safety potential; crash analysis; rear-end crashes; pedestrian crashes; intersection crashes.
Shift control of vehicle automatic transmission based on traffic congestion identification
by Guang Xia, You Zheng, Xiwen Tang, Baoqun Sun, Shaojie Wang
Abstract: This work builds a T-S fuzzy neural network that identifies traffic congestion conditions by using average vehicle speed, average throttle opening and frequency of brake pedal actuation as evaluation factors. A strategy that controls the shift of vehicle automatic transmission based on the identified congestion conditions is also devised. This strategy divides the vehicle automatic transmission system into the upper identification and decision-making layer and the lower shift execution layer. Simulation and real vehicle tests are performed to verify the effectiveness of the proposed strategy. The results show that congestion conditions can be accurately identified by using the T-S fuzzy neural network and that the proposed layered correction shift control strategy can prevent the frequent changing of gears under congestion conditions, thereby reducing the wear of the shift execution parts and the braking system.
Keywords: traffic congestion identification; T-S fuzzy neural network; shift correction; layered control; real vehicle test.
Motion planning methods for autonomous vehicles in disordered traffic systems: a comparative analysis and future research directions
by Caleb Ronald Munigety
Abstract: Most works that exist in the field of autonomous vehicle motion planning are in the context of orderly or lane-oriented traffic. However, the traffic conditions that exist in many countries like India, China, Bangladesh, etc. are disordered where the vehicles move anywhere laterally without complying with the lane discipline rule. This characteristic makes the task of planning by an autonomous vehicle manoeuvring in disordered traffic streams highly complex and computationally rigorous. The motion planning problem of autonomous vehicles in traffic systems where lane-discipline is not necessarily followed bears a resemblance with the mobile-robot motion planning. Thus, this paper reviews various techniques available for motion planning of robots in conjunction with the issue of motion planning of autonomous vehicles in disordered traffic conditions, highlights the advantages and limitations of different approaches alongside simulation-based comparative analysis, and finally puts forth some research directions.
Keywords: motion planning; mobile robot; autonomous vehicle; disordered traffic; non-lane discipline.
Ceres: the wheeled mobile agricultural robot designed for vegetables care
by H. Guillermo Sanchez, Leonardo E. Solaque-Guzmán, Adriana Riveros Guevara
Abstract: Outdoor robots require high performance and safety, in the agricultural robot case, factors such as temperature, light conditions or obstacles, must be taken into account. However, despite the navigation problem, the robot must support agricultural tasks during the navigation through the crop, these agricultural tasks are highly related to the specific crop. In this paper, an agricultural robot (Ceres) for vegetable crops is described. The design of controllers for velocities and path following based on classical techniques and Lyapunov theory are shown. Likewise, agricultural tasks such as fumigating, fertilising and weeding are performed by Ceres, the implemented tasks are made using different tools coupled to a movable 3D system to act over the entire crop during the robot operation. The systems named above are described throughout the document. All subsystems are integrated using the Robotic Operating System (ROS). Finally, field test are performed to the entire robot over grass terrain.
Keywords: mobile robot; vegetables care; agricultural robot; modelling; navigation; path following; path planning; controller design; agricultural task; robotic operating system; Lyapunov-based controller.
Special Issue on: Advancements, Applications and Challenges in Internet of Vehicles for Smart Transportation
Socio-realistic optimal path planning for indoor real-time autonomous mobile robot navigation
by Rahul Shivaji Pol, B. Sheela Rani, M. Murugan
Abstract: An autonomous mobile robotic navigation system consists of many modules which work co-ordinately and concurrently. The most important module is the realistic and optimal path planning algorithm (ROPPA) through which the overall system improves it performance. Many algorithms have been developed and deployed for data structures, and computer games are partially modified for use in real time robotic environments. The major drawback of such modified algorithms is they are designed for unconstrained artificial environments where the robot's collision with a static obstacle or moving object is partially allowed. Many researchers successfully developed the path planning algorithm through improving the basic A* algorithm, such as D* lite, theta*, any angle path planning, and jump point search. In real environments one should seek the optimal path along with maintaining uniform safer distance with the objects or in-path obstacles. This paper describes the implementation and evaluation of a new realistic optimal path planning algorithm which follows the safer distance rule through exploring minimum workspace arena. Along with less memory overhead, the algorithm explores shortest final path with fewer sub-paths if they exist. The experimentation with different map sizes and obstacle densities clearly defines improvement in ROPPA over the other path planning methods.
Keywords: grid-based segmentation; A*; theta*; any angle path planning; optimal path planning; safer path planning; realistic optimal path planning.
Towards an immersive and safer driving experience using computer vision integrated with encoded vibro-tactile feedback
by Rajshekhar Mukherjee, Dharmendra Kumar Mahato, Sangeeta Yadav, Amit Pundir, Geetika Jain Saxena
Abstract: This paper claims to set up an immersive, responsive vehicle driving system and mechanism for an assisted driving technology. The purpose is to expand the sensory horizon of humans while driving and is motivated by absence of any such system in real world. The system can control and direct an assembly of electronic devices in real time, through usage of an image acquisition subsystem, an object-recognition and tracking algorithm and a haptic modelling subsystem working in-tandem with the user. The object tracking subsystem operates in real time to determine the current position of a vehicle in front by using a camera and continuously updates it in a live video feed, while also identifying and tracking the moving or stationary vehicle. The haptic system, which is integrated with the tracking system, has been programmed to warn the driver of the potential threats that moving/stationary vehicles may generate. All the subsystems are updated and synchronised with each other in real-time to produce a seamless and smooth transition between frames, facilitating a precise and immersive driving experience for anyone. The high accuracy and robustness of the proposed system makes it a versatile component, which can be integrated in variety of applications for enhancing a person's reality perception.
Keywords: object tracking; image detection; SURF; vibro-tactile; vision-to-touch; human-centred computing; immersive reality.
Special Issue on: Embedded Autonomous Architecture and Its Connectivity for Automotives
A hybrid approach to perform test case prioritisation and reduction for software product line testing
by Satendra Kumar, Raj Kumar, Mohit Mittal
Abstract: Software Product Line (SPL) is a popular research area in software engineering which deals with numerous products generated simultaneously. There is requirement for a tool or framework that can test all the products. Owing to an increase in the number of products there is an exponential increase in the number of features. SPL implies Test Case Prioritization (TCP) and Test Case Reduction (TCR) techniques to alleviate the problem of testing. This paper proposes a hybrid approach which combined K-means and Principal Component Analysis (PCA) approaches to perform SPL testing. The hybrid approach provides the solution to SPL testing using K-means clustering and PCA. We use Euclidean distance equation to perform K-means clustering. PCA helps in reducing the dimension of data from multidimensional to two dimensional (2D) data, and as a result, it can plot the clusters using K-means clustering. The experimental results show that hybrid approach provides better results than the random, similarity, and ICPL algorithm order the APFD metric. Our proposed approach reduces the test cases to the minimum amount and it can also be used to perform efficient SPL testing that not only improves the effectiveness and efficiency of fault detection but also improves the effectiveness of Test Suite Reduction (TSR). Thus this hybrid approach can be useful to alleviate the effort of SPL tester
Keywords: feature model; software product line testing; test case prioritisation; test case reduction; principal component analysis; K-means clustering.
Design and modelling of a hybrid fuel cell and sloar-based electric vehicle
by Ravipati Srikanth, M. Venkatesan
Abstract: Design and modelling of a hybrid electric vehicle using two different energy sources (fuel cell and photovoltaic array) are presented in this paper. In this proposed model, a High Gain Interleaved Boost Converter (HGIBC) is used to extract the maximum power from the PV array and fuel cell. To optimise the maximum power, a single MPPT system has been applied to both the fuel cell stack and the solar panel. A Brushless DC Motor (BLDC) is used to evaluate the performance of the electric vehicle. A BLDC motor has been designed to coerce the hybrid electric vehicle using a three phase inverter. The performance characteristics of the system have been analysed in terms of rise time, peak overshoot and efficiency. Finally, the simulation results are verified through MATLAB/Simulink.
Keywords: hybrid electric vehicle; HGIBC; fuel cell; photovoltaic array; brushless DC motor; MATLAB/Simulink.