使用深度学习在低计算量人形机器人上进行实时的目标检测和识别
Real-Time Object Detection and Recognition on Low-Compute Humanoid Robots using Deep Learning
摘要：
我们设想，在不久的将来，人形机器人将共享家庭空间，并通过对象操作来帮助我们进行日常和常规活动。需要为机器人开发的基本技术之一是使它们能够检测物体并识别它们以进行有效的操作，并对涉及这些物体的物体做出实时的【形】决策。在本文中，我们描述了一种新的架构，使多个低计算NAO机器人执行实时的【形】检测，识别和定位的对象在其相机视图，并采取可编程的行动的基础上检测到的对象。所提出的目标检测和定位算法是YOLOv3的经验修改，基于多场景的室内实验，具有更小的权重大小和更少的计算需求。权重的量化和重新调整卷积的滤波器大小和层安排改善了机器人相机馈送的低分辨率图像的推理时间。YOLOv3是在对限界框算法进行比较研究后选择的，目的是选择一种在信息保留、低推理时间和高精度之间取得完美平衡的算法，以实现实时的【形】目标检测和定位。该架构还包括一个有效的端到端流水线，用于将实时的【形】帧从相机馈送到神经网络，并使用其结果来指导机器人执行与检测到的类标签相对应的可定制动作。
英文摘要：
We envision that in the near future, humanoid robots would share home space and assist us in our daily and routine activities through object manipulations. One of the fundamental technologies that need to be developed for robots is to enable them to detect objects and recognize them for effective manipulations and take real-time decisions involving those objects. In this paper, we describe a novel architecture that enables multiple low-compute NAO robots to perform real-time detection, recognition and localization of objects in its camera view and take programmable actions based on the detected objects. The proposed algorithm for object detection and localization is an empirical modification of YOLOv3, based on indoor experiments in multiple scenarios, with a smaller weight size and lesser computational requirements. Quantization of the weights and re-adjusting filter sizes and layer arrangements for convolutions improved the inference time for low-resolution images from the robot s camera feed. YOLOv3 was chosen after a comparative study of bounding box algorithms was performed with an objective to choose one that strikes the perfect balance among information retention, low inference time and high accuracy for real-time object detection and localization. The architecture also comprises of an effective end-to-end pipeline to feed the real-time frames from the camera feed to the neural net and use its results for guiding the robot with customizable actions corresponding to the detected class labels.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
