摘　　要

隨著科學(xué)技術(shù)的不斷進(jìn)步，工業(yè)機(jī)器人相關(guān)技術(shù)也在不斷向前發(fā)展，控制系統(tǒng)作為機(jī)器人的大腦已成為工業(yè)機(jī)器人領(lǐng)域的一個熱門研究方向。為此，本文重點(diǎn)研究了控制系統(tǒng)中的軌跡規(guī)劃部分，使用五次均勻B樣條函數(shù)插值關(guān)鍵路徑點(diǎn)，同時使用混合遺傳算法處理機(jī)器人軌跡的相關(guān)約束，最終實(shí)現(xiàn)了兼顧時間最優(yōu)與沖擊最優(yōu)的二次軌跡規(guī)劃方案。
首先，本文全面總結(jié)了國內(nèi)外有關(guān)軌跡規(guī)劃的研究成果，詳細(xì)分析了三種常見的軌跡規(guī)劃優(yōu)化方案——最優(yōu)時間軌跡規(guī)劃、最優(yōu)能量軌跡規(guī)劃和最優(yōu)沖擊軌跡規(guī)劃的優(yōu)缺點(diǎn)。最優(yōu)時間軌跡規(guī)劃沒有考慮到?jīng)_擊因素的影響，對沖擊還能作進(jìn)一步優(yōu)化，因此本文提出了一種二次軌跡規(guī)劃方案，即先通過最優(yōu)時間軌跡規(guī)劃得到最小執(zhí)行時間，然后在最小執(zhí)行時間內(nèi)進(jìn)行最優(yōu)沖擊軌跡規(guī)劃，進(jìn)而規(guī)劃出一條既高效又平滑的運(yùn)動軌跡。同時考慮到最優(yōu)軌跡規(guī)劃比較耗時，因此采用離線規(guī)劃在線插值的策略。
然后，本文使用MATLAB配合Robotics Toolbox工具箱對二次軌跡規(guī)劃方案進(jìn)行了離線仿真。由于五次均勻B樣條曲線不僅能夠保證了各關(guān)節(jié)速度和加速度連續(xù)性還能保證了各關(guān)節(jié)沖擊的連續(xù)性，連續(xù)平滑的沖擊可以減少機(jī)械振動，延長機(jī)器人的工作壽命，因此本文選用B樣條函數(shù)軌跡規(guī)劃的插值函數(shù)。由于普通的遺傳算法只能有效的處理不等式約束，對等式約束的處理效果不好，而罰函數(shù)法卻能夠很好的處理等式約束，所以本文結(jié)合兩種優(yōu)化算法的優(yōu)點(diǎn)提出一種混合遺傳算法，有效解決了軌跡規(guī)劃中運(yùn)動學(xué)與動力學(xué)約束。選用PUMA560為對象進(jìn)行仿真與實(shí)驗(yàn)，結(jié)果表明，該方案可以獲得比較理想的機(jī)器人運(yùn)動軌跡，所提出的混合遺傳算法能有效提高全局尋優(yōu)的性能和算法運(yùn)行的穩(wěn)定性。
最后，本文在TI的OMAP3530平臺上進(jìn)行在線插值。對OMAP3530平臺的軟件架構(gòu)進(jìn)行了深入研究，掌握了從ARM端調(diào)用DSP端插值算法的整個流程，即使用Codec Engine進(jìn)行算法調(diào)用，DSPLink負(fù)責(zé)數(shù)據(jù)在CMEM分配的共享緩存中交互，整個過程還需LPM對DSP進(jìn)行電源管理。

關(guān)鍵字 工業(yè)機(jī)器人；軌跡規(guī)劃；混合遺傳算法；B樣條；OMAP3530




























Abstract

With the improvement of science and technology, the industrial robot is also developed quickly, and control system as the robot’s brain has become a popular research direction in the field of industrial robot. Therefore, this paper mainly focuses on the research of trajectory planning in the control system by adopting fifth-order uniform B-splines to interpolate a sequence of critical path points and using the hybrid genetic algorithm to deal with the constraints of robot’s trajectory, and finally the quadratic trajectory planning which achieves the time-optimal and the jerk-optimal is implemented.
First of all, the research results of the trajectory planning at home and broad is summarized in this paper, and the advantages and disadvantages of three optimization method of trajectory planning - the time-optimal trajectory planning, the energy-optimal trajectory planning and the jerk-optimal trajectory planning are analyzed in detail. The time-optimal trajectory planning doesn't take into account the effect of the jerk. In order to do further optimization of the jerk, a quadratic trajectory planning method is put forward in this paper. That is to say, the trajectory with minimum execution time is obtained by using time-optimal trajectory planning, then an efficient and smooth trajectory is generated in this minimum execution time by using jerk-optimal trajectory planning. Moreover, the strategy of planning offline and interpolation online is used in this paper.
Then, the quadratic trajectory planning method is simulated offline by using MATLAB and Robotics Toolbox. The velocity, acceleration and jerk of each joint trajectorys interpolated by B-splines are continuous, and the continuous and smooth jerk can reduce mechanical vibration and also extend the working life of the robot, so this paper chooses fifth-order uniform B-splines as the interpolation function. Genetic algorithm can effectively deal with inequality constraints, but it has bad effects on equality constraints. the penalty function is a good algorithm to solve the problem of equality constraints. So a hybrid genetic algorithm is put forward in this paper by combining the advantages of the two kinds of optimization algorithms to resolve the kinematics and dynamics constraints of trajectory planning. The method is applied to PUMA560, the result of simulation and experiment shows that the method can provide ideal trajectory and the hybrid genetic algorithm is more efficient and stable.
Finally, the interpolation algorithm is running on the TI OMAP3530 platform. The software of OMAP3530 is researched in depth in this paper. The entire process of calling interpolation algorithm from the ARM to the DSP is well knew. Codec Engine is used to call algorithm from the ARM to the DSP. DSPLink is responsible for data exchange in the Shared buffer distributed by CMEM, and the whole process also need LPM to manage power supply of the DSP.

Keywords industrial robot；trajectory planning；hybrid genetic algorithm；B-Spline；OMAP3530






















目 錄
摘要 I
Abstract Ⅲ-b..