Full 4D Change Analysis of Topographic Point Cloud Time Series using Kalman Filtering

4D topographic point cloud data contain information on surface change processes and their spatial and temporal characteristics, such as the duration, location, and extent of mass movements, e.g., rockfalls or debris flows. To automatically extract and analyse change and activity patterns from this data, methods considering the spatial and temporal properties are required. The commonly used M3C2 point cloud distance reduces uncertainty through spatial averaging for bitemporal analysis. To extend this concept into the full 4D domain, we use a Kalman filter for point cloud change analysis. The filter incorporates M3C2 distances together with uncertainties obtained through error propagation as Bayesian priors in a dynamic model. The Kalman filter yields a smoothed estimate of the change time series for each spatial location, again associated with an uncertainty. Through the temporal smoothing, the Kalman filter uncertainty is, in general, lower than the individual bitemporal uncertainties, which therefore allows detection of more change as significant. In our example time series of bi-hourly terrestrial laser scanning point clouds of around 6 days (71 epochs) showcasing a rockfall-affected high-mountain slope in Tyrol, Austria, we are able to almost double the number of points where change is deemed significant (from 14.9 % to 28.6 % of the area of interest). Since the Kalman filter allows interpolation and, under certain constraints, also extrapolation of the time series, the estimated change values can be temporally resampled. This can be critical for subsequent analyses that are unable to deal with missing data, as may be caused by, e.g., foggy or rainy weather conditions. We demonstrate two different clustering approaches, transforming the 4D data into 2D map visualisations that can be easily interpreted by analysts. By comparison to two state-of-the-art 4D point cloud change methods, we highlight the main advantage of our method to be the extraction of a smoothed best estimate time series for change at each location. A main disadvantage of not being able to detect spatially overlapping change objects in a single pass remains. In conclusion, the consideration of combined temporal and spatial data enables a notable reduction in the associated uncertainty of the quantified change value for each point in space and time, in turn allowing the extraction of more information from the 4D point cloud dataset.

Permitted speed decision of single-unit trucks with emergency braking maneuver on horizontal curves under rainy weather

Under adverse weather conditions, visibility and the available pavement friction are reduced. The improper selection of speed on curved road sections leads to an unreasonable distribution of longitudinal and lateral friction, which is likely to cause rear-end collisions and lateral instability accidents. This study considers the combined braking and turning maneuvers to obtain the permitted vehicle speed under rainy conditions. First, a braking distance computation model was established by simplifying the relationship curve between brake pedal force, vehicle braking deceleration, and braking time. Different from the visibility commonly used in the meteorological field, this paper defines "driver’s sight distance based on real road scenarios" as a threshold to measure the longitudinal safety of the vehicle. Furthermore, the lateral friction and rollover margin is defined to characterize the vehicle’s lateral stability. The corresponding relationship between rainfall intensity-water film thickness-road friction is established to better predict the safe speed based on the information issued by the weather station. It should be noted that since the road friction factor of the wet pavement not only determined the safe vehicle speed but also be determined by the vehicle speed, so we adopt Ferrari’s method to solve the quartic equation about permitted vehicle speed. Finally, the braking and turning maneuvers are considered comprehensively based on the principle of friction ellipse. The results of the TruckSim simulation show that for a single-unit truck, running at the computed permitted speed, both lateral and longitudinal stability meet the requirements. The proposed permitted vehicle speed model on horizontal curves can provide driving guidance for drivers on curves under rainy weather or as a decision-making basis for road managers.

Thermal Analysis of Tilapia Fish Drying by Hybrid Solar Thermal Drying System

The present work presents a hybrid solar thermal drying of Tilapia fish to improve the product quality and satisfy the importers. The developed hybrid dryer utilized direct solar drying, a solar air heater and a thermal backup unit which sustains the drying process during the night, cloudy and rainy weather conditions. Besides, a new feature of the developed dryer utilizes the flue gas exhausted from the thermal unit to enhance the updraft in the drying chamber by re-injection of the flue gases in the chimney. The initial moisture content of the Tilapia fish used in the investigation was 246.6% on a dry basis, equivalent to 74% on a wet basis. The investigations were repeated three times on different days. Experimental results showed that the moisture content was reduced to an average final of 17.0% db (5.0% wb) within 17.5 hours, while in the open sun drying, it required around 48-72 hours. Hybrid solar drying required around 72% shorter time than open sun drying. The average overall drying efficiency of the developed system for drying Tilapia fish was 13.0%. The Re-injection technique used in the present hybrid solar-thermal system has excluded the need for an electric source for air extraction from the drying chamber, which is highly desired in the rural and fishery regions.

Behind-The-Scenes (BTS): Wiper-Occlusion Canceling for Advanced Driver Assistance Systems in Adverse Rain Environments

Driving in an adverse rain environment is a crucial challenge for vision-based advanced driver assistance systems (ADAS) in the automotive industry. The vehicle windshield wiper removes adherent raindrops that cause distorted images from in-vehicle frontal view cameras, but, additionally, it causes an occlusion that can hinder visibility at the same time. The wiper-occlusion causes erroneous judgments by vision-based applications and endangers safety. This study proposes behind-the-scenes (BTS) that detects and removes wiper-occlusion in real-time image inputs under rainy weather conditions. The pixel-wise wiper masks are detected by high-pass filtering to predict the optical flow of a sequential image pair. We fine-tuned a deep learning-based optical flow model with a synthesized dataset, which was generated with pseudo-ground truth wiper masks and flows using auto-labeling with acquired real rainy images. A typical optical flow dataset with static synthetic objects is synthesized with real fast-moving objects to enhance data diversity. We annotated wiper masks and scenes as detection ground truths from the collected real images for evaluation. BTS outperforms by achieving a 0.962 SSIM and 91.6% F1 score in wiper mask detection and 88.3% F1 score in wiper image detection. Consequently, BTS enhanced the performance of vision-based image restoration and object detection applications by canceling occlusions and demonstrated it potential role in improving ADAS under rainy weather conditions.

Analysing pedestrian–vehicle conflicts behaviour at urban pedestrian crossings

Pedestrian safety has become a serious problem with the rapid growth of motorised vehicle in transportation system in developing counties. Pedestrians often respond differently to changes in surrounding and traffic conditions. A study was undertaken to investigate pedestrians’ gap acceptance and the parameters affecting their risk-taking behaviours based on time-to-collision and post-encroachment-time indexes. Three signalised intersections and two midblock crossings were selected in Qazvin, Iran. A total of 752 pedestrians were examined by video recording and field observation, and pedestrians’ gap acceptance behaviour was estimated by using binary logit model. Results showed that the average time to collision and post-encroachment time were 4.27 s and 1.44 s, respectively. In addition, the presence of children alongside the older pedestrians led to a less risk-taking crossing. Additionally, pedestrian risk-taking was reduced by increasing both time indexes. Rainy weather also reduced pedestrians’ risk-taking behaviour. Elasticity analysis indicated that parameters such as pedestrians’ conflict with vehicles at the first or second half of the crossings, walking with a child, speed of the approaching vehicle, the crossing type and running while crossing were the most important factors in pedestrian risk-taking.

Analysis of the Effect of the Speed Factor on Highway Safety Using the Machine Learning Method

Speed is one of the most important factors that can significantly change the severity of accidents. Providing a model with predictive factors leads to designing traffic plans to promote safety. This study aims to create statistical models for accidents occurred on Firuzkuh highway, Iran. Moreover, the probability of each type of accident was determined using the logit model. Various modeling methods, such as backward, forward, and entering methods, were evaluated to find the best method. Finally, since the backward method had the best performance in terms of R2 and goodness of fit, the logit model of accidents was created. According to the model, the independent variables of the 12-24 hours, rainy weather, a speed of 81-95 and 96-110 km/h, the lack of attention ahead and the Pride brand of vehicle increased the severity of accidents, while the variables with negative coefficients of Tuesdays, the summer and spring seasons, sunny weather, a male driver, and daylight, reduced the severity of accidents.

Performance Test of Solar – Powered Ice Maker Machine: Case Study in Bantul, Yogyakarta

Research on performance test of a solar-powered ice maker machine has been conducted in Bantul, Yogyakarta. This study aimed to observe the correlation between intensity of sunlight to the power battery capacity rates generated from solar panels in regard with the performance of ice maker machine. The testing was conducted during various weather conditions i.e. sunny, cloudy and light rain. The type of ice maker observed was a flake ice maker machine with specifications of the production capacity of 105 - 120 kgs/day, producing flake ice with dimensions of 2 x 3 x 3 mm3. The energy of the machine was generated by nine solar panels with maximum power of 200 Wp (watt peak) per panel. A set of three panels was arranged in series, it was then coupled to other sets in parallel. The results showed that the power battery capacity was in corresponded to the sunlight intensity during sunny weather with the correlation: y = 0.009x - 26.08, while during cloudy dan raining conditions, the power capacity rates of the battery showed a declining with the correlation: y = 0.008x - 23.92 and y = 0.007x + 69.41, respectively. The ice production capacity during sunny, cloudy and light rainy weathers were 4.34 kg ice/hour; 4.63 kg ice / hour and 4.17 kg ice / hour respectively. Input power from solar panels depends on the intensity of sunlight. The ice produced by ice makers in cloudy weather conditions is much greater than the ice produced during sunny or rainy weather conditions.

Lightweight Object Detection Ensemble Framework for Autonomous Vehicles in Challenging Weather Conditions

The computer vision systems driving autonomous vehicles are judged by their ability to detect objects and obstacles in the vicinity of the vehicle in diverse environments. Enhancing this ability of a self-driving car to distinguish between the elements of its environment under adverse conditions is an important challenge in computer vision. For example, poor weather conditions like fog and rain lead to image corruption which can cause a drastic drop in object detection (OD) performance. The primary navigation of autonomous vehicles depends on the effectiveness of the image processing techniques applied to the data collected from various visual sensors. Therefore, it is essential to develop the capability to detect objects like vehicles and pedestrians under challenging conditions such as like unpleasant weather. Ensembling multiple baseline deep learning models under different voting strategies for object detection and utilizing data augmentation to boost the models’ performance is proposed to solve this problem. The data augmentation technique is particularly useful and works with limited training data for OD applications. Furthermore, using the baseline models significantly speeds up the OD process as compared to the custom models due to transfer learning. Therefore, the ensembling approach can be highly effective in resource-constrained devices deployed for autonomous vehicles in uncertain weather conditions. The applied techniques demonstrated an increase in accuracy over the baseline models and were able to identify objects from the images captured in the adverse foggy and rainy weather conditions. The applied techniques demonstrated an increase in accuracy over the baseline models and reached 32.75% mean average precision (mAP) and 52.56% average precision (AP) in detecting cars in the adverse fog and rain weather conditions present in the dataset. The effectiveness of multiple voting strategies for bounding box predictions on the dataset is also demonstrated. These strategies help increase the explainability of object detection in autonomous systems and improve the performance of the ensemble techniques over the baseline models.