| Autor: |
L K Pavithra, B. Rajesh Kanna, A V Shreyas Madhav |
| Rok vydání: |
2021 |
| Předmět: |
|
| Zdroj: |
2021 5th International Conference on Computer, Communication and Signal Processing (ICCCSP). |
| DOI: |
10.1109/icccsp52374.2021.9465538 |
| Popis: |
Simultaneous Localization and Mapping plays an integral role in the field of autonomous robotics for ensuring proper navigation and exploration in unknown environments. Applications of SLAM range from surveillance and monitoring to self driving cars or UAVs. The high computational complexity involved in simultaneously tracking the trajectory of the robot while constantly updating the surrounding map based on every scan observation is disadvantageous to the mainstream adoption of SLAM. With the advent of parallel computing and multicore architectures, serial programs and algorithms can be accelerated in terms of execution time and efficiency. This paper explores the effects of CPU thread parallelism on 2D LiDAR SLAM in an indoor environment. The recorded scans are matched and loop closure detection is performed. The final trajectory and mapping is finetuned with the help of pose graph optimization. The number of threads and CPU cores are varied to draw analysis of their effects on the serial algorithm. The simulated results show a performance increase and a decrease of 19.1% in execution time for the parallelized algorithm utilizing 4 CPUs, compared to its serial version. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
