機器人系統的智慧型感測器設計及電池殘電量的估測

Abstract

隨著科技的快速發展與進步，機器人系統也逐漸被受到重視。感測器設計的技術與電池殘電量的估測在監控機器人系統中扮演著重要的角色。因此，本論文主要的目的，在發展能夠即時準確監控機器人運動訊號之系統 (如電壓、電流、溫度、角速度及殘電量等)並且應用在本實驗室開發的人型機器人。 為了研發出一具備上述功能之即時監控的感測系統應用於人形機器人，本論文主要可分為兩大研究主題。第一部份將著重於微機電陀螺儀及其感測電路設計與無線溫度感測器設計。在微機電陀螺儀的研究中，頻率匹配是一個很重要的議題，本文設計一具有解耦合結構之單框架微機電陀螺儀(Single-gimbaled Decoupled Gyroscope, SGDG)，並利用所提出之回授頻率匹配演算法(Feedback Frequency-Matching Algorithm, FFMA)，改善因頻率不匹配造成的訊號衰減，並可利用此方法結合正交誤差消除法(Quadrature Error Cancellation, QEC)，減少因製程變異造成機械結構不完美所產生的正交誤差。模擬結果顯示，FFMA與QEC皆可以提高陀螺儀的效能。其次，本論文設計出一個具高線性及高靈敏度的交換電容式陀螺儀感測電路，此一電路架構對寄生電容不敏感。因此，非常適合於陀螺儀的量測，並且可以利用訊號處理將偏差電壓消除。在溫度感測器開發部分，由於之前的設計為根據CMOS正比於絕對溫度(Proportional-To-Absolute Temperature, PTAT)原理，對於量測溫度範圍僅從20度到120度，且該溫度感測器僅提供95%的線性度與2.3 mV/°C的敏感度。因此，為了改善之前設計的量測範圍、線性度及靈敏度，本文設計一個新的CMOS無線感測器晶片，根據CMOS元件具有雙零溫度係數點(Double Zero Temperature Coefficient, DZTC)的特性，此晶片具有兩個電壓参考源、一個電流参考源以及一個溫度感測器。量測範圍從−20度到120度，該溫度感測器可提供97%的高線性度與9.55 mV/°C的高敏感度，此值高於之前設計的感測器4.15倍之多。此晶片亦內建一個八位元逐步逼近式(Successive-Approximation-Register, SAR)類比數位轉換器與433 MHz無線傳輸。此外，為了加快電路及感測器佈局的速度，本文也同時設計一個以基因演算法為基礎的自動佈置系統(Automatic Placement System, APS)來改善佈局設計的效率。 論文第二部分，則提出一個以基因演算法為基礎的電池封裝最佳化設計(Optimization of the Battery Pack, OBP)，並且滿足電池平衡及可防止電池過充或過放。實驗結果顯示此方法可以大大減低溫度差異和功率損失以及精確地估測電池殘電量。其次，本論文也提出一個新的電池殘電量估測的整合方法(Modified ECE + EKF)。此方法考慮了自我放電、溫度及電池殘電量對於庫倫效率的影響因素，並結合擴充卡爾曼濾波器(Extended Kalman Filter, EKF)修正的殘電量初始值使其快速收斂至正確值。實驗結果顯示所提出的方法優於其他傳統的演算法，且精確的估測電池殘電量在1%內。

With the advance of science and technology, there has been increasing interest in robotic systems. Two of the most important requirements for monitoring such systems are sensor design technology and battery state of charge (SOC) estimation. This dissertation aims to develop an integrated monitoring system that can accurately monitor motion signals (including voltage, current, temperature, angular velocity, and SOC) in real-time and can be applied to a humanoid robot developed by our laboratory. The dissertation discusses the development of this integrated system in two major parts. The first focuses on developing intelligent sensors, including the MEMS gyroscope, its sensing circuit, and the wireless temperature sensor. In MEMS gyroscope research, frequency matching is an important issue. The single-gimbaled decoupled gyroscope (SGDG) is presented first, and the proposed feedback frequency-matching algorithm (FFMA) is used to improve signal attenuation caused by frequency mismatch. The FFMA combined quadrature error cancellation algorithm (QEC) can also reduce quadrature error caused by imperfection in the mechanical structure. According to the simulation results, both FFMA and QEC can increase gyroscope performance. Secondly, an ultra linear and high sensitivity switched-capacitor sensing circuit, with parasitic-insensitive topology, can be realized at the same time. This circuit is very suitable for gyroscope measurement and can use signal processing to cancel the voltage offset. As for the wireless temperature sensor, in our previous design, based on the CMOS proportional-to-absolute temperature (PTAT) principle, the temperature sensor can only has the sensitivity of 2.3 mV/°C with linearity up to 95% for the temperature range from 20 °C to 120 °C. In order to improve the measurement range, linearity, and sensitivity of our previous design using the PTAT principle, a combined device based on the principle of CMOS double zero temperature coefficient (DZTC) points is first created at the chip level with two voltage references, one current reference, and one temperature sensor. The results show that the chip can achieve linearity up to 97% with a sensitivity of 9.55 mV/°C, in a wide temperature range from −20 °C to 120 °C. The proposed temperature sensor has 4.15-times better sensitivity than the previous design. An 8-bit successive-approximation-register (SAR) ADC and a 433 MHz wireless transmitter are also integrated in this chip. In order to improve the efficiency of analog IC layout design, the automatic placement system (APS), which is expected to improve the layout design procedure for analog IC product development, is proposed to quickly provide a layout design placement. The second part of the dissertation first proposes a method based on a genetic algorithm (GA) for optimization of the battery pack (OBP). The proposed method considers the cell balance of battery packs and thus avoids cell over-discharge and over-charge. Validation results indicate that the proposed method can greatly reduce temperature variation and power loss due to temperature effect, and can accurately estimate battery packs’ SOC and terminal voltage. Secondly, a new SOC estimation method, “Modified ECE + EKF,” is proposed, combining a modified ECE method with an EKF-based method. The modified ECE method considers self-discharge, influence of temperature, and SOC on coulombic efficiency, while the EKF-based method is used to make the approximate initial SOC value converge to its real value. The experimental results show that the proposed method is superior to several traditional techniques, giving a SOC estimation accuracy within 1% of the true value.