Overcoming Challenges of Applying Reinforcement Learning for Intelligent Vehicle Control

Abstract

Reinforcement learning (RL) is a booming area in artificial intelligence. The applications of RL are endless nowadays, ranging from fields such as medicine or finance to manufacturing or the gaming industry. Although multiple works argue that RL can be key to a great part of intelligent vehicle control related problems, there are many practical problems that need to be addressed, such as safety related problems that can result from non-optimal training in RL. For instance, for an RL agent to be effective it should first cover all the situations during training that it may face later. This is often difficult when applied to the real-world. In this work we investigate the impact of RL applied to the context of intelligent vehicle control. We analyse the implications of RL in path planning tasks and we discuss two possible approaches to overcome the gap between the theorical developments of RL and its practical applications. Specifically, firstly this paper discusses the role of Curriculum Learning (CL) to structure the learning process of intelligent vehicle control in a gradual way. The results show how CL can play an important role in training agents in such context. Secondly, we discuss a method of transferring RL policies from simulation to reality in order to make the agent experience situations in simulation, so it knows how to react to them in reality. For that, we use Arduino Yún controlled robots as our platforms. The results enhance the effectiveness of the presented approach and show how RL policies can be transferred from simulation to reality even when the platforms are resource limited.

A Comprehensive Overview of Classical and Modern Route Planning Algorithms for Self-Driving Mobile Robots

“…The increased number of nodes traversed reduces the algorithm's efficiency. To identify the most efficient route across the network, this approach employs graph searching techniques [9]. To approximate the configuration space, several discrete cellgrid spaces and lattices are used [14].…”

“…When it comes to providing precise and high-quality answers to optimization and discovery issues, genetic algorithms are widely recognised as one of the most regularly used optimization methods [1,2,4,5]. With no previous knowledge of what may be the optimal answer to the issue, GA is inspired by the natural selection notion [8][9][10][11][12][13][14]. Evolutionary operators like mutation, crossover, and selection are used to acquire data from the world before determining the best response for a specific circumstance.…”

“…As a function of a path point in the neural network, it defines energy [33]. Depending on the location of the route point, the network follows the energydepleted path [9]. Finally, the least total energy path is found.…”

“…[1], [5], [6], [9], [11], [13], [14], [15], [38] Heuristic A-Star/A* Algorithm Created in accordance with Dijkstra's algorithm.…”

“…When it comes to laying out a route, there are three types: traditional algorithms, bionic algorithms, and reinforcementbased path planning algorithms [9]. Campbell et al classified path planning into the classical and heuristic approach [2].…”