Adaptive Motion Planning with Artificial Potential Fields Using a Prior Path

Motion planning in an autonomous agent is responsible for providing smooth, safe and efficient navigation. Many solutions for dealing this problem have been offered, one of which is, Artificial Potential Fields (APF). APF is a simple and computationally low cost method which keeps the robot away from the obstacles in environment. However, this approach suffers from trapping in local minima of potential function and then fails to produce motion plans. Furthermore, Oscillation in presence of obstacles or in narrow passages is another disadvantage of the method which makes it unqualified for many planning problems. In this paper we aim to resolve these deficiencies by a novel approach which employs a prior path between origin and goal configuration of the robot. Therefore, the planner guarantees to lead the robot to goal area while the inherent advantages of potential fields remain. For path planning stage, we intend to use randomized sampling methods such as Rapidly-exploring Random Trees (RRT) or its derivatives, however, any path planning approach can be utilized. We have also designed an optimization procedure for evolving the motion plans towards optimal solution. Then genetic algorithm is applied to find smoother, safer and shorter plans. In our experiments, we apply a simulated vehicle in Webots simulator to test and evaluate the motion planner. Our experiments showed our method to enjoy improving the performance and speed in comparison to basic approaches.

Node-Adapt, Path-Adapt and Tree-Adapt:Model-Transfer Domain Adaptation for Random Forest

Random Forest (RF) is a successful paradigm for learning classifiers due to its ability to learn from large feature spaces and seamlessly integrate multi-class classification, as well as the achieved accuracy and processing efficiency. However, as many other classifiers, RF requires domain adaptation (DA) provided that there is a mismatch between the training (source) and testing (target) domains which provokes classification degradation. Consequently, different RF-DA methods have been proposed, which not only require target-domain samples but revisiting the source-domain ones, too. As novelty, we propose three inherently different methods (Node-Adapt, Path-Adapt and Tree-Adapt) that only require the learned source-domain RF and a relatively few target-domain samples for DA, i.e. source-domain samples do not need to be available. To assess the performance of our proposals we focus on image-based object detection, using the pedestrian detection problem as challenging proof-of-concept. Moreover, we use the RF with expert nodes because it is a competitive patch-based pedestrian model. We test our Node-, Path- and Tree-Adapt methods in standard benchmarks, showing that DA is largely achieved.