Mohamed Kharbeche

An optimization-based heuristic for the robotic cell problem

This study investigates an optimization-based heuristic for the robotic cell problem. This problem arises in automated cells and is a complex flow shop problem with a single transportation robot and a blocking constraint. We propose an approximate decomposition algorithm. The proposed approach breaks the problem into two scheduling problems that are solved sequentially: a flow shop problem with additional constraints (blocking and transportation times) and a single machine problem with precedence constraints, time lags, and setup times. For each of these problems, we propose an exact branch-and-bound algorithm. Also, we describe a genetic algorithm that includes, as a mutation operator, a local search procedure. We report the results of a computational study that provides evidence that the proposed optimization-based approach delivers high-quality solutions and consistently outperforms the genetic algorithm. However, the genetic algorithm delivers reasonably good solutions while requiring significantly shorter CPU times.

Real-time Driver Drowsiness Detection for Android Application Using Deep Neural Networks Techniques

Road crashes and related forms of accidents are a common cause of injury and death among the human population. According to 2015 data from the World Health Organization, road traffic injuries resulted in approximately 1.25 million deaths worldwide, i.e. approximately every 25 seconds an individual will experience a fatal crash. While the cost of traffic accidents in Europe is estimated at around 160 billion Euros, driver drowsiness accounts for approximately 100,000 accidents per year in the United States alone as reported by The American National Highway Traffic Safety Administration (NHTSA). In this paper, a novel approach towards real-time drowsiness detection is proposed. This approach is based on a deep learning method that can be implemented on Android applications with high accuracy. The main contribution of this work is the compression of heavy baseline model to a lightweight model. Moreover, minimal network structure is designed based on facial landmark key point detection to recognize whether the driver is drowsy. The proposed model is able to achieve an accuracy of more than 80%.

Pedestrians' Crossing Behavior at Marked Crosswalks on Channelized Right-Turn Lanes at Intersections

Abstract: Unsignalized marked crosswalks are problematic locations from pedestrians’ safety perspective. Although the priority rule at such locations is clear; pedestrians have the absolute right of way over vehicles, driver often compete with pedestrians over the right of way which risks pedestrian safety and impose extra delays on pedestrians. In developing countries, as concluded in many previous studies, vehicles usually do not give right of way to pedestrians, leaving them with the only choice to wait until an accepted gap is available. In Gulf Cooperation Council (GCC) countries where vehicles are the predominant mode of travel, pedestrians are receiving lesser priority. Drivers usually hijack the right of way from pedestrians which often causes safety threats. Therefore, pedestrians loose the reason for crossing at these designed locations pushing them to cross at arbitrary locations increasing their safety risk. This paper investigates influencing factors on the crossing behavior of pedestrians at marked crosswalks located on dedicated right-turn lanes. A study site from Doha, Qatar was selected for video recording and data analysis. A sample of 235 pedestrian observations was used for waiting behavior, crossing speed, accepted gaps, and driver yielding behavior analysis. The results showed that the waiting behavior was independent of pedestrian characteristics and relied only on the traffic characteristics. In addition, the average crossing speed was 1.43m/s and the gender, distraction, and group significantly affected the crossing speed. Beside, the distracted pedestrians and pedestrians crossing in groups accepted significantly larger gaps compared to undistracted and individual pedestrians. Moreover, about 15% of drivers yielded for pedestrians, yielding was irrespective of gender and mainly affected by the crossing direction. Consequently, innovative strategies in terms of engineering measures and awareness are needed to improve pedestrian safety at these locations.

A Free-Slack-Based Genetic Algorithm for the Robotic Cell Problem with Controllable Processing Times

We present a novel genetic algorithm for the Robotic Cell Problem with controllable processing times. This challenging problem arises in an automated production cell that consists of m consecutive machines as well as a material handling robot. The problem requires finding the operations processing times, job assignment, and robot movements. The objective is to minimize the makespan subject to a budget constraint. We describe a free-slack-based genetic algorithm for the linear resource consumption case. We present the results of a computational study and we provide evidence that the proposed algorithm consistently outperforms MIP-based heuristics from the literature.

An optimization-based heuristic for the machine reassignment problem

We address the machine reassignment problem proposed in the context of the ROADEF/EURO challenge 2012 in partnership with Google. The problem consists in reassigning a set of processes to a set of multiple-resource machines so as to minimize a weighted function of the machines load, the resources balance, and the costs of moving processes while satisfying numerous constraints. We propose an optimization-based heuristic that requires decomposing the problem into a sequence of small-sized instances that are iteratively solved using a general MIP solver. To speed-up the solution process several algorithmic expedients are embedded. Extensive computational experiments provide evidence that the proposed approach exhibits a very good performance.

Minimizing Expected Attacking Cost in Networks

A branch-and-bound algorithm is devised to determine the optimal attack strategy to disconnect a network where the objective is to minimize the expected attacking cost. The attacker cannot launch an attack if its cost is beyond his available budget or its probability of success falls below a threshold level. The proposed branch-andbound algorithm includes, among other features, a dynamic programming-based lower bound as well as a preprocessing algorithm which aims at identifying unattackable links and removing irrelevant ones. Extensive use of the min-cut algorithm is made to derive valid upper bounds and to perform feasibility tests. Preliminary numerical implementation shows potential to provide exact solutions for medium-sized networks within reasonable time.

Prediction of Traffic Conflicts at Signalized Intersections using SSAM

Abstract The use of microsimulation to model the vehicles movement and pedestrian movements within a traffic network is widely undertaken to test and evaluate operational performance of a traffic network under different traffic conditions and control schemes. However, few studies have used microsimulation techniques to study pedestrian-vehicle interactions and potential conflicts, as safety assessment tool. This paper demonstrates the use of microsimulation environment to predict vehicle-vehicle and pedestrian-vehicle conflicts at signalized intersections. A case study from Doha in the State of Qatar was used as a study site. The real-life conflicts were observed and recorded, along with traffic and pedestrians’ data. The studied intersection is then modeled and calibrated using VISSIM microsimulation tool, where vehicles and pedestrians’ trajectories were generated. Then, Surrogate Safety Assessment Model (SSAM) was used to analyze the simulated trajectories to identify potential conflicts within the study area. The results showed that potential conflicts could be reasonably predicted. Moreover, microsimulation can be used to predict the location of potential conflicts while scenario testing and the results can be determined to assess the impact of geometric improvement in reducing potential conflicts.

Pedestrians’ Speed Analysis at Signalized Crosswalks

Abstract Pedestrian speed is essential for designing signal timings as well as for understanding pedestrian safety issues. This paper presents analysis of three types of pedestrian speeds (entry speed, crossing speed, and exit speed) at three signalized crosswalks in the State of Qatar. Pedestrian movements were tracked using TrafficAnalyzer software and data were analyzed to determine the effect of signal indications, crosswalk length, and crossing direction on pedestrian speeds. The results of this analysis showed that only 23.69% pedestrians crossed legally, during pedestrian green or pedestrian flashing green interval. Moreover, the pedestrian entry speeds were significantly affected by the pedestrian signal indications. Furthermore, the crossing speeds were positively correlated with crosswalk length for pedestrians crossing on green and red indications while pedestrian exit speeds were independent of crosswalk length but significantly affected by crossing direction.

Throughput optimization for the Robotic Cell Problem with Controllable Processing Times

In this paper, we present a MIP-based heuristic and an effective genetic algorithm for the Robotic Cell Problem with Controllable Processing Times (RCPCPT). This problem arises in modern automated manufacturing systems and requires simultaneously scheduling jobs, machines, and transportation devices in order to maximize the throughput or minimize the makespan. The RCPCPT is modeled as a flow shop problem with blocking constraints, a single transport robot, and controllable processing times. This latter feature of the model refers to the fact that the processing times are not fixed but vary linearly with the acceleration cost and therefore should be determined as part of the problem output. We formulate the problem as a nonlinear mixed-integer programming formulation and we use its linearized form to derive LP- and MIP-based heuristics. In addition, we proposed a genetic algorithm consistently yields near-optimal solution and it encompasses several novel features including, an original solution encoding as well as a mutation operator that requires iteratively solving MIPs in order to generate feasible processing times. Finally, we present a computational study for the proposed formulation, heuristics and genetic algorithm and we provide an empirical evidence of the effectiveness of the MIP-based heuristic for small instances and the genetic algorithm for large instances.

An overview of pedestrian signal setting and implementation in the State of Qatar

Abstract: Pedestrian signal control is a main influencing factor on the Level of Service of crossing facilities and on the safety performance considering conflicts with vehicles. In signal design, pedestrian clearance time is a key design parameter for ensuring safe pedestrian crossing at signalized crosswalks. The performance of pedestrian signal control and its impact on pedestrian behavior in the Gulf Cooperation Council (GCC) Countries is rarely addressed in the literature. The characteristics of population, cultural diversity, as well as the extreme hot weather conditions, may lead to significantly different pedestrian behavior in terms of crossing manoeuvres (path and speed), compliance with signal control and interaction with vehicular traffic. This study reviews the pedestrian signal design practices in various countries and investigates the current signal settings along with their adequacy in the State of Qatar using empirical observations. The empirical analysis showed that the 85 th percentile crossing times were longer than the provided Pedestrian Flashing Green (PFG) intervals at the observed crosswalks. Study sites are characterized with unrealistically long Buffer Intervals (BIs) which, from one side, provide sufficient time for pedestrians who started crossing during PFG to complete the crossing. On the other side, they encourage pedestrians to continue crossing after the end of PFG. Additionally, the speed analysis indicated that the observed 15th percentile speed was 1.22 m/s which is similar to the assumed design speed by the Qatar Traffic Control Manual QTCM (2015). Further, the analysis showed that pedestrian crossing speed during PFG or BI was significantly higher than that during PG.