Multi-sensor fusion for Automated Driving: Selecting model and optimizing on Embedded platform

Jagannathan, Shyam; Mody, Mihir; Jones, J. P.; Swami, Pramod; Poddar, Deepak

doi:10.2352/issn.2470-1173.2018.17.avm-256

Cited by 7 publications

(7 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These emerging phenomena have prompted researchers to explore new possibilities for sophisticated smart devices [ 1 ] to be embedded in various real objects and to cope with various environmental changes. Recently, multiple similar and/or dissimilar sensors, as shown in Figure 1 , have been widely used to provide precise sensing information from different viewpoints [ 2 , 3 ] and to realize the Internet of Things in a cyber-physical system [ 4 ]. Given that the accuracy of a sensor system is dictated by the degree to which repeated measurements under unchanging conditions are able to produce the same results, a multi-sensor system has typically been thought of as a way to guarantee the accuracy of a measurement system [ 5 ].…”

Section: Introductionmentioning

confidence: 99%

Hybrid Particle Swarm Optimization for Multi-Sensor Data Fusion

Kim

Suh

2018

Sensors

View full text Add to dashboard Cite

A hybrid particle swarm optimization (PSO), able to overcome the large-scale nonlinearity or heavily correlation in the data fusion model of multiple sensing information, is proposed in this paper. In recent smart convergence technology, multiple similar and/or dissimilar sensors are widely used to support precisely sensing information from different perspectives, and these are integrated with data fusion algorithms to get synergistic effects. However, the construction of the data fusion model is not trivial because of difficulties to meet under the restricted conditions of a multi-sensor system such as its limited options for deploying sensors and nonlinear characteristics, or correlation errors of multiple sensors. This paper presents a hybrid PSO to facilitate the construction of robust data fusion model based on neural network while ensuring the balance between exploration and exploitation. The performance of the proposed model was evaluated by benchmarks composed of representative datasets. The well-optimized data fusion model is expected to provide an enhancement in the synergistic accuracy.

show abstract

Section: Introductionmentioning

confidence: 99%

Hybrid Particle Swarm Optimization for Multi-Sensor Data Fusion

Kim

Suh

2018

Sensors

View full text Add to dashboard Cite

show abstract

“…In the multiple-point object process in Table 3, Kalman filter-based methods are the most popular among model-based approaches (Jagannathan et al, 2018). The methods are classified by what type of model is used and what estimation technique is applied.…”

Section: High-levelmentioning

confidence: 99%

Automated Driving Vehicles

Song

2021

Vehicle Dynamics

View full text Add to dashboard Cite

This chapter introduces a full scope of autonomous vehicle technology ranging from perception to control. While a literature review for commercial environment sensors and sensor fusion to both model-based and data-driven approaches is summarized in Sect. 2, detailed decision and motion planning strategies for urban driving situations as well as highway are described. Moreover, some experimental results are demonstrated in Sect. 3. In Sects. 4 and 5, lateral and longitudinal control methods are proposed respectively in the hierarchical structure. The several upper controllers are designed to track the desired path and their performances are compared experimentally. The torque controller for a steering actuator as a lower controller is designed based on nonlinear control techniques. Similarly, the upper and lower controller for longitudinal control with consideration of human driver characteristics is proposed for better naturalistic driving performance. Finally, a scenario-based approach for validation of the autonomous vehicle technology including all proposed algorithms is investigated in the end.

show abstract

“…Some literature focused on LiDAR + camera multi-sensor approach [1,2,7] and a single fusion method. Other literature focused on camera + radar + LiDAR multi-sensor fusion using state estimators i.e., Kalman filters or particle filters [8,9,10]. Also, some works focused on the fusion method techniques without any real-world implementation or any specific set of sensors.…”

Section: Fusion Systems Literature Reviewmentioning

confidence: 99%

“…In [8], a fusion system is proposed based on a 2D laser scanner and camera sensors with Unscented Kalman filter (UKF) and joint probabilistic data association to detect vehicles. Reference [10] proposes a choice of fusion techniques using a Kalman filter, given a sensor configuration and its model parameters. Reference [9] investigated the problem of estimating wheelchair states in an indoor environment using UKF to fuse sensor data.…”

Section: Fusion Systems Literature Reviewmentioning

confidence: 99%

Real-Time Hybrid Multi-Sensor Fusion Framework for Perception in Autonomous Vehicles

Jahromi

Tulabandhula

Cetin

2019

Sensors

133

View full text Add to dashboard Cite

There are many sensor fusion frameworks proposed in the literature using different sensors and fusion methods combinations and configurations. More focus has been on improving the accuracy performance; however, the implementation feasibility of these frameworks in an autonomous vehicle is less explored. Some fusion architectures can perform very well in lab conditions using powerful computational resources; however, in real-world applications, they cannot be implemented in an embedded edge computer due to their high cost and computational need. We propose a new hybrid multi-sensor fusion pipeline configuration that performs environment perception for autonomous vehicles such as road segmentation, obstacle detection, and tracking. This fusion framework uses a proposed encoder-decoder based Fully Convolutional Neural Network (FCNx) and a traditional Extended Kalman Filter (EKF) nonlinear state estimator method. It also uses a configuration of camera, LiDAR, and radar sensors that are best suited for each fusion method. The goal of this hybrid framework is to provide a cost-effective, lightweight, modular, and robust (in case of a sensor failure) fusion system solution. It uses FCNx algorithm that improve road detection accuracy compared to benchmark models while maintaining real-time efficiency that can be used in an autonomous vehicle embedded computer. Tested on over 3K road scenes, our fusion algorithm shows better performance in various environment scenarios compared to baseline benchmark networks. Moreover, the algorithm is implemented in a vehicle and tested using actual sensor data collected from a vehicle, performing real-time environment perception.

show abstract

Multi-sensor fusion for Automated Driving: Selecting model and optimizing on Embedded platform

Cited by 7 publications

References 0 publications

Hybrid Particle Swarm Optimization for Multi-Sensor Data Fusion

Hybrid Particle Swarm Optimization for Multi-Sensor Data Fusion

Automated Driving Vehicles

Real-Time Hybrid Multi-Sensor Fusion Framework for Perception in Autonomous Vehicles

Contact Info

Product

Resources

About