Object detection is recognized as one of the most critical research areas for the perception of self-driving cars. Current vision systems combine visible imaging, LIDAR, and/or RADAR technology, allowing perception of the vehicle’s surroundings. However, harsh weather conditions mitigate the performances of these systems. Under these circumstances, thermal imaging becomes the complementary solution to current systems not only because it makes it possible to detect and recognize the environment in the most extreme conditions, but also because thermal images are compatible with detection and recognition algorithms, such as those based on artificial neural networks. In this paper, an analysis of the resilience of thermal sensors in very unfavorable fog conditions is presented. The goal was to study the operational limits, i.e., the very degraded fog situation beyond which a thermal camera becomes unreliable. For the analysis, the mean pixel intensity and the contrast were used as indicators. Results showed that the angle of view (AOV) of a thermal camera is a determining parameter for object detection in foggy conditions. Additionally, results show that cameras with AOVs 18° and 30° are suitable for object detection, even under thick fog conditions (from 13 m meteorological optical range). These results were extended using object detection software, with which it is shown that, for the pedestrian, a detection rate ≥90% was achieved using the images from the 18° and 30° cameras.

Methods based on 64-beam LiDAR can provide very precise 3D object detection. However, highly accurate LiDAR sensors are extremely costly: a 64-beam model can cost approximately USD 75,000. We previously proposed SLS–Fusion (sparse LiDAR and stereo fusion) to fuse low-cost four-beam LiDAR with stereo cameras that outperform most advanced stereo–LiDAR fusion methods. In this paper, and according to the number of LiDAR beams used, we analyzed how the stereo and LiDAR sensors contributed to the performance of the SLS–Fusion model for 3D object detection. Data coming from the stereo camera play a significant role in the fusion model. However, it is necessary to quantify this contribution and identify the variations in such a contribution with respect to the number of LiDAR beams used inside the model. Thus, to evaluate the roles of the parts of the SLS–Fusion network that represent LiDAR and stereo camera architectures, we propose dividing the model into two independent decoder networks. The results of this study show that—starting from four beams—increasing the number of LiDAR beams has no significant impact on the SLS–Fusion performance. The presented results can guide the design decisions by practitioners.

Low cost and small size integrated sensors bring a significant interest to redundant sensing systems such as sensor array systems. Redundancy may permit to increase both performance and dependability of individual sensors at the system level, however, the emergence of these redundant systems has generated the need for dynamic algorithms able to detect and act against the presence of errors in system elements. Then, a new adaptive algorithm for sensor array systems is presented in this work. This approach is generic and can be implemented for different types of sensor systems, besides it does not require any information about the inputs. The proposed real-time algorithm is based on the following: 1) estimation of sensor individual variances at each time step, 2) identification and removal of sensors affected by errors, and 3) reincorporation of sensors after recovery or replacement. Variance estimation is revisiting MInimum Norm Quadratic Unbiased Estimation (MINQUE) method to allow real-time operation and requires that the number of sensors is strictly greater than two times the number of signals to be measured. Consequently, the proposed method is able to support the presence of m-1 faulty devices, where m is the total number of sensors less two times the number of measured signals. Performance of the proposed algorithm is demonstrated through theoretical demonstrations, simulations, and experimental results.

Sensors have become crucial elements in many consumer, industrial and high-end applications. Then, system-level simulations of multi-sensors system require realistic behavioral models of each sensor. The proposed generic model allows implementing the complete behavior of a sensor including uncertainties and tolerances, nonlinearities, various noises and so on. This sensor model can be customized by extracting information from a datasheet, a power spectral density graph or an Allan deviation graph.

Implementations of multisensor systems such as sensor array systems and sensor network systems have increased in the last years. This has created the need for algorithms to detect and act against errors and faults in the elements that make up the system. In this work, the particular case of redundant similar-sensor systems is addressed. Here, a new method for dynamically estimating the variance of all sensors in the system is presented. This method is used together with an online weighted average algorithm for estimating and correcting confidence degrees of information coming from sensors in a multisensor system. Precision in the individual variance estimation is related to the number of sensors, making it suitable for applications with a large number of sensors. Also, variance estimator has low computational cost, which permits its implementation on smart devices such as based-MEMS systems.