Lidar-Monocular Visual Odometry with Genetic Algorithm for Parameter Optimization

Abstract

Lidar-Monocular Visual Odometry (LIMO), a odometry estimation algorithm, combines camera and LIght Detection And Ranging sensor (LIDAR) for visual localization by tracking camera features as well as features from LIDAR measurements, and it estimates the motion using Bundle Adjustment based on robust key frames. For rejecting the outliers, LIMO uses semantic labelling and weights of the vegetation landmarks. A drawback of LIMO as well as many other odometry estimation algorithms is that it has many parameters that need to be manually adjusted according to the dynamic changes in the environment in order to decrease the translational errors. In this paper, we present and argue the use of Genetic Algorithm to optimize parameters with reference to LIMO and maximize LIMO's localization and motion estimation performance. We evaluate our approach on the well known KITTI odometry dataset and show that the genetic algorithm helps LIMO to reduce translation error in different datasets.

Automatic Parameter Optimization Using Genetic Algorithm in Deep Reinforcement Learning for Robotic Manipulation Tasks

“…Binary coding with concatenated parameters is used to encode chromosomes. [19] shows one such example of GA paired with Lidar-monocular visual odometry (LIMO).…”

“…[8], [9], [10], and [11] are some of the closely related works. These publications' findings add to the growing body of data that using a GA to automatically modify the hyper-parameters for DRL can greatly enhance efficiency.…”

GA-DRL: Genetic Algorithm-Based Function Optimizer in Deep Reinforcement Learning for Robotic Manipulation Tasks

“…Some of the closely related work includes [8], [9] and [10]. The results from these papers provide more evidence that when a GA is used to automatically tune the hyperparameters for DRL, efficiency can be largely improved.…”

Deep Learning Hyperparameter Optimization for Breast Mass Detection in Mammograms

“…Even though the performance of the whole image classifier to detect a mass in a mammogram has been improved, the authors [27] do not justify the use of many constant parameters used in transferring the learning to a dataset with minimal/no ROI information. [23] and [24] have shown that automatic tuning of such parameters can greatly enhance the performance of the overall system. [23] made use of GA with Deep Reinforcement Learning (GA-DRL), and [24] used GA with Lidar-monocular visual odometry (GA-LIMO).…”

“…[23] and [24] have shown that automatic tuning of such parameters can greatly enhance the performance of the overall system. [23] made use of GA with Deep Reinforcement Learning (GA-DRL), and [24] used GA with Lidar-monocular visual odometry (GA-LIMO). Both of these works proved that when GA is used for parameter tuning, it has the potential of producing encouraging results.…”