具人機互動能力之自走式仿人類眼球運動機器人

Abstract

本計劃嘗試結合仿人類眼球運動機器人以及自走車，成為自走式仿人眼機器 人，以視覺為基礎執行人機互動。目前計畫主持人已經開發以下技術與系統： 1. 仿人類眼球運動機器人機構設計。 2. 人臉偵測技術。 3. 物體追蹤技術。 4. 灰預測技術。 5. 運動描述系統。 本計劃將藉由偵測人臉、手勢、固定顏色物體，設計人機互動模式，讓機器人保 持適當的互動距離，以類神經網路為基礎學習描述手勢，學習協調雙眼與頸部的 運動，利用灰預測理論加強追蹤能力，此外也採用基因演算法搜尋最佳的探索路 徑規畫。預計一年內可達成以下之目標： 1. 製作自走式仿人眼機器人。 2. 利用人臉偵測保持互動距離。 3. 手勢描述系統。 4. 協調雙眼與頸部運動。 5. 應用灰預測於物體追蹤。 6. 設計具有最佳的探索路徑的搜尋模式。 完成以上目標之自走式仿人眼機器人將擁有人機互動的能力，能辨識多種手勢， 也可自主地在環境中探索。

This project proposes a mobile humanoid eye-ball robot, or MHER for simplicity, constructed from combining a humanoid eye-ball robot and a mobile robot. The MHER is designed to interact with human based on vision processing. Recently, some related key technologies have been devloped and listed as below: 1. Machanism design of humanoid eye-ball robot. 2. Face detection technology. 3. Object tracking technology. 4. Grey prediction technology. 5. Motion description system. Based on the detection of face, gesture, and object with predefined color, this project will develop human-robot interaction, able to detect the interactive distance, recognize the gesture based on recurrent neural networks, coordinate the movements of eyes and neck, and enhance tracking ability by grey predictor. In addition, the genetic algorithm will be adopted in the search of optimal path for exploration. This project would take one year to achieve the following goals: 1. Implementation of the MHER. 2. Detection of interactive distance. 3. Implementation of gesture description system. 4. Coordination of the movement of eyeballs and neck. 5. Tracking technology based on grey predictor. 6. Design of the optimal path for exploration. After this project, the developed MHER could interact with human beings, describe several gestures, and autonomously explore the environment.