索控式中央驱动上肢康复机器人

目的根据居家康复的要求与趋势,针对国内外上肢康复机器人体积庞大、传动链复杂以及驱动噪声大的缺点,设计了一款索控式中央驱动上肢康复机器人。方法首先采用钢丝绳和同步带组成的传动系统进行动力传递,并采用中央驱动式设计,即将电机驱动设备统一安置于远离患者的底座上;然后通过Solidworks建立三维模型,并设计动力传递系统的总体方案;最后通过Solidworks中的Motion模块对机器人肩关节屈曲/伸展运动、肩关节内收/外展运动、肘关节屈曲/伸展运动以及肩肘关节联动运动进行运动学仿真,验证所设计的上肢康复机器人机械臂结构的合理性及设定轨迹的可行性。结果肩关节、肘关节以及腕关节运动学仿真曲线证实了该上肢康复机器人能够平稳地完成规划的轨迹动作,从而验证了机械结构设计合理,方案可行。结论索控式中央驱动上肢康复机器人具有动力传递系统轻便以及结构简单的优点,可以协助患者完成各种康复运动。     

基于Simulink的上肢康复训练人机整体建模与分析

机器人康复训练是解决中风瘫痪患者康复训练需求的重要方法,康复机器人设定的训练参数对患者能力的适应程度是决定训练能否加快患者康复进程的重要因素。目前临床上的康复训练内容由康复医师对患者进行量表评估并结合自己的经验制定,这种经验式的训练方法在康复机器人训练过程中无法实现。为了探讨上肢康复机器人运动训练参数与患者运动能力之间的关系,本文建立了牵引式上肢康复机器人训练的Simulink人机整体模型,并将模型计算的人体肩关节、肘关节运动与实际动作下的肩关节、肘关节运动进行了对照。分析显示仿真结果与实验数据存在明显的相关性,从而证明了模型的准确性。进一步地,本文根据模型仿真结果拟合了人机接触端平面画圆半径与人体肩关节、肘关节主动运动自由度的线性函数关系,为临床上康复机器人制定训练目标提供了量化参考。     

基于多传感器信息的新型穿戴式上肢外骨骼康复机器人

目的 设计基于多传感器信息的新型穿戴式上肢外骨骼康复机器人,以解决上肢外骨骼康复机器人便携性不佳、患者参与度较低、训练模式自适应不足等问题,并探究受试者穿戴外骨骼时肌肉激活程度、肌电信号预测关节角度的准确性以及实现上肢康复训练的可行性.方法 该设备机械结构包括肘关节和腕关节,采用模块化设计并结合3D打印技术;控制系统包括肌电采集、应力采集、姿态采集等单元,并设计主动、被动和助动三种训练模式.受试者穿戴外骨骼机器人后进行屈-伸肘实验,对比有、无辅助力时手臂肌肉激活程度;分析肘关节角度,并对比肌电信号预测的关节运动角度;验证机器人运行性能与应力检测效果.结果 受试者穿戴外骨骼康复机器人安全可靠地完成了屈-伸肘动作,受试者肱二头肌、肱三头肌肌肉激活程度在有、无辅助力时分别减弱约32％、11％,肌电信号预测关节角度准确度约95％,应力测量值误差均低于5％.结论 上肢外骨骼机器人可以给人体提供辅助力、预测关节角度,机器人通过肌电、应力以及位置信息辅助患者实现上肢康复训练具有可行性.     

一种上肢康复机器人机械臂重力平衡方法

目的针对现有上肢康复机器人机械臂质量较大且其自体质量无法忽略等问题,基于重力平衡原理提出一种上肢康复机器人机械臂重力平衡方法。方法通过角度传感器检测肩关节屈/伸角度,将经过滤波的模拟量信号传送到单片机,单片机计算得到目标力矩,并作为控制参数发送到驱动器,驱动器驱动电机进入力矩模式并输出力矩,平衡上肢康复机器人机械臂重力。结果实验所用机械臂长度L_1=300 mm、L_2=300 mm,关节运动角度0°～140°,机械臂自体质量M_1=3.0 kg、M_2=4.4 kg;电机可提供最大力矩27 N·m、最大电流0.766 A;满足实验需求。在肩关节屈/伸角度范围为30°～120°的机械臂上选取7组角度进行实验。实验证明,在30°～75°内实际输出力矩大于理论输出力矩,机械臂保持平衡且有向上运动的趋势;在90°～120°内,实际输出力矩略小于理论力矩,机械臂保持平衡且有向下运动的趋势。结论实验证明,笔者设计能很好地实现机械臂重力平衡。     

基于光学运动捕捉的上肢康复机器人可用性测试研究

目的对一款上肢康复机器人进行可用性测试研究,验证其安全性、有效性和舒适性,并提出改进设计的意见。方法运用光学式运动捕捉系统采集受试者进行康复训练时的角度数据,再对受试者进行主观问卷调查。结果快速上肢评估量表结果显示上肢康复机器人配置相对合适,但需要进一步深入研究。角度数据在肩关节内收外展运动和肘关节屈伸运动上表现出显著相关性,而在腕关节数据的统计结果上表现出无相关性。主观问卷调查总得分表明上肢康复机器人舒适性总体上是令人满意的,但在手臂固定装置的舒适性、训练过程中肌肉拉伸程度和座位高度设置的问题上存在设计不合理。结论运用光学式运动捕捉系统可以定量评价上肢康复机器人的性能,对康复机器人的设计及改进具有重要意义。     

基于伺服电机的上肢康复机器人力矩 交互控制系统

目的为了实现上肢康复机器人主动训练中力矩控制的精确性,设计一种基于伺服电机的上肢康复机器人力矩交互控制系统。方法首先利用三自由度中央驱动式上肢康复机器人实验平台建立由运动意图采集模块和伺服力矩控制模块组成的力矩交互系统;再通过建立上肢动力学模型提出上肢助力训练算法;最后通过力矩响应实验、运动意图检测实验。结果依据在一定电流输入范围内,输出力矩能够保持稳定,验证了基于直流伺服电机实现助力训练方法的可行性。采用伺服力矩控制模块在目标力矩设定后,输出力矩在1 ms时间内能够达到设定的目标值并能够保持稳定,目标力矩响应的实时性良好,由此可以得出力矩交互控制方法达到了人机力矩交互稳定的结果。结论采用上肢康复机器人力矩交互控制方法可以在主动训练中实现较为精确的力矩控制。     

人服系统上肢交互生物力学仿真模型

目的通过建立人-舱外服上肢交互生物力学仿真模型计算穿着舱外服后航天员上肢关节力矩和肌肉力,满足出舱活动风险评估的需求。方法分别建立舱外服手臂的刚体运动学模型和关节阻尼力矩迟滞模型以描述舱外服关节的运动和力学特性。通过对舱外航天服肘部和人体肘部位置进行约束实现人体和舱外服手臂之间的运动学耦合,利用虚拟反作用力元实现两者之间的动力学耦合,在反向运动生物力学架构下建立一体化仿真模型。利用该模型对宇航员穿着加压、未加压舱外服和不穿着舱外服3种工况下肘弯曲/伸展进行仿真案例分析。结果3种工况下肱二头肌的预测肌肉激活和积分肌电的相关性分别为0.86、0.71、0.65,肱三头肌对应的相关性分别为0.75、0.61、0.60,采用预测肌肉激活和积分肌电的一致性定性验证了模型的正确性,利用舱外服肘关节阻尼力矩与人体肘关节肌肉承受力矩之间的一致性验证了模型的合理性。结论该人服系统上肢交互生物力学仿真模型能有效计算航天员穿着舱外服后的上肢关节力矩和肌肉力,且仿真结果和实验表明,加压后舱外服关节阻尼力矩造成较大的人体关节力矩和肌肉负荷,为航天员出舱活动中的体力负荷和骨肌风险评估提供方法学支撑。     

人服系统上肢交互生物力学仿真模型

目的 通过建立人-舱外服上肢交互生物力学仿真模型计算穿着舱外服后航天员上肢关节力矩和肌肉力，满足出舱活动风险评估的需求。方法 分别建立舱外服手臂的刚体运动学模型和关节阻尼力矩迟滞模型以描述舱外服关节的运动和力学特性。通过对舱外航天服肘部和人体肘部位置进行约束实现人体和舱外服手臂之间的运动学耦合，利用虚拟反作用力元实现两者之间的动力学耦合，在反向运动生物力学架构下建立一体化仿真模型。利用该模型对宇航员穿着加压、未加压舱外服和不穿着舱外服3种工况下肘弯曲/伸展进行仿真案例分析。结果 3种工况下肱二头肌的预测肌肉激活和积分肌电的相关性分别为0.86、0.71、0.65，肱三头肌对应的相关性分别为0.75、0.61、0.60，采用预测肌肉激活和积分肌电的一致性定性验证了模型的正确性，利用舱外服肘关节阻尼力矩与人体肘关节肌肉承受力矩之间的一致性验证了模型的合理性。结论 该人服系统上肢交互生物力学仿真模型能有效计算航天员穿着舱外服后的上肢关节力矩和肌肉力，且仿真结果和实验表明，加压后舱外服关节阻尼力矩造成较大的人体关节力矩和肌肉负荷，为航天员出舱活动中的体力负荷和骨肌风险评估提供方法学支撑。     

人服系统上肢交互生物力学仿真模型

目的 通过建立人-舱外服上肢交互生物力学仿真模型计算穿着舱外服后航天员上肢关节力矩和肌肉力,满足出舱活动风险评估的需求。方法 分别建立舱外服手臂的刚体运动学模型和关节阻尼力矩迟滞模型以描述舱外服关节的运动和力学特性。通过对舱外航天服肘部和人体肘部位置进行约束实现人体和舱外服手臂之间的运动学耦合,利用虚拟反作用力元实现两者之间的动力学耦合,在反向运动生物力学架构下建立一体化仿真模型。利用该模型对宇航员穿着加压、未加压舱外服和不穿着舱外服3种工况下肘弯曲/伸展进行仿真案例分析。结果 3种工况下肱二头肌的预测肌肉激活和积分肌电的相关性分别为0.86、0.71、0.65,肱三头肌对应的相关性分别为0.75、0.61、0.60,采用预测肌肉激活和积分肌电的一致性定性验证了模型的正确性,利用舱外服肘关节阻尼力矩与人体肘关节肌肉承受力矩之间的一致性验证了模型的合理性。结论 该人服系统上肢交互生物力学仿真模型能有效计算航天员穿着舱外服后的上肢关节力矩和肌肉力,且仿真结果和实验表明,加压后舱外服关节阻尼力矩造成较大的人体关节力矩和肌肉负荷,为航天员出舱活动中的体力负荷和骨肌风险评估提供方法学支撑。     

中央驱动式上肢康复机器人运动学建模与分析

为在结构上减小机械臂的体积,同时还减小电机噪音、辐射等不良因素对上肢功能障碍患者的影响,本文提出了上肢康复机器人的中央驱动式传动结构。利用Denavit-Hartenberg(D-H)表示法对中央驱动式上肢康复机器人进行运动学建模,得到了其正、逆运动学方程。使用SolidWorks三维建模软件进行三维建模,并让其在预设康复轨迹下进行运动仿真实验,得到了各关节的角度-时间曲线及机械臂手柄的位置-时间曲线。实验结果表明,机械臂手柄能在设定的康复轨迹下运动,验证了中央驱动式结构设计的合理性。同时,由机械臂手柄位置信息通过逆运动学方程计算得到的各关节的角度-时间曲线与从实验中得到的曲线最大误差不到3°,验证了所得运动学方程的有效性。     

Deliberate utilization of interaction torques brakes elbow extension in a fast throwing motion

We tested the hypothesis that in fast arm movements the CNS deliberately utilizes interaction torques to decelerate (brake) joint rotations. Twelve subjects performed fast 2-D overarm throws in which large elbow extension velocities occurred. Joint motions were computed from recordings made with search coils; joint torques were calculated using inverse dynamics. After ball release, a large follow-through shoulder extension acceleration occurred that was initiated by shoulder extensor muscle torque. This shoulder acceleration produced a flexor interaction torque at the elbow that initiated elbow deceleration (braking). An instantaneous mechanical interaction of passive torques then occurred between elbow and shoulder, i.e., elbow extension deceleration produced a large shoulder extensor interaction torque that contributed to the shoulder extension acceleration which, simultaneously, produced a large elbow flexor interaction torque that contributed to elbow extension deceleration, and so on. Late elbow flexor muscle torque also contributed to elbow deceleration. The interaction of passive torques between shoulder and elbow was braked by shoulder flexor muscle torque. In this mechanism, shoulder musculature contributed to braking elbow extension in two ways: shoulder extensors initiated the mechanical interaction of passive torques between shoulder and elbow and shoulder flexors dissipated kinetic energy from elbow braking. It is concluded that, in fast 2-D throws, the CNS deliberately utilizes powerful interaction torques between shoulder and elbow to brake motion at the elbow.     

A novel shoulder–elbow mechanism for increasing speed in a multijoint arm movement

The speed of arm movements is normally increased by increasing agonist muscle activity, but in overarm throwing, an additional effect on speed may come from exploitation of interaction torques (a passive torque associated with motion at adjacent joints). We investigated how the central nervous system (CNS) controls interaction torques at the shoulder and elbow to increase speed in 2-D overarm throwing. Twelve experienced throwers made slow, medium, and fast 2-D throws in a parasagittal plane. Joint motions were computed from recordings made with search coils; joint torques were calculated using inverse dynamics. For slow and medium-speed throws, elbow extension was primarily produced by elbow muscle torque. For fast throws, there was an additional late-occurring elbow extensor interaction torque. Parceling out this elbow extension interaction torque revealed that it primarily arose from shoulder extension deceleration. Surprisingly, shoulder deceleration before ball release was not caused by shoulder flexor (antagonist) muscle torque. Rather, shoulder deceleration was produced by passive elbow-to-shoulder interaction torques that were primarily associated with elbow extension acceleration and velocity. It is concluded that when generating fast 2-D throws, the CNS utilized the arm’s biomechanical properties to increase ball speed. It did this by coordinating shoulder and elbow motions such that an instantaneous mechanical positive feedback occurred of interaction torques between shoulder and elbow before ball release. To what extent this mechanism is utilized in other fast multijoint arm movements remains to be determined.     

Position-dependent torque coupling and associated muscle activation in the hemiparetic upper extremity

Previous studies have demonstrated abnormal joint torque coupling and associated muscle coactivations of the upper extremity in individuals with unilateral stroke. We investigated the effect of upper limb configuration on the expression of the well-documented patterns of shoulder abduction/elbow flexion and shoulder adduction/elbow extension. Maximal isometric shoulder and elbow torques were measured in stroke subjects in four different arm configurations. Additionally, an isometric combined torque task was completed where subjects were required to maintain various levels of shoulder abduction/adduction torque while attempting to maximize elbow flexion or extension torque. The dominant abduction/elbow flexion pattern was insensitive to changes in limb configuration while the elbow extension component of the adduction/extension pattern changed to elbow flexion at smaller shoulder abduction angles. This effect was not present in control subjects without stroke. The reversal of the torque-coupling pattern could not be explained by mechanical factors such as muscle length changes or muscle strength imbalances across the elbow joint. Potential neural mechanisms underlying the sensitivity of the adduction/elbow extension pattern to different somatosensory input resultant from changes in limb configuration are discussed along with the implications for future research.     

Compensation for interaction torques during single- and multijoint limb movement

During multijoint limb movements such as reaching, rotational forces arise at one joint due to the motions of limb segments about other joints. We report the results of three experiments in which we assessed the extent to which control signals to muscles are adjusted to counteract these "interaction torques." Human subjects performed single- and multijoint pointing movements involving shoulder and elbow motion, and movement parameters related to the magnitude and direction of interaction torques were manipulated systematically. We examined electromyographic (EMG) activity of shoulder and elbow muscles and, specifically, the relationship between EMG activity and joint interaction torque. A first set of experiments examined single-joint movements. During both single-joint elbow (experiment 1) and shoulder (experiment 2) movements, phasic EMG activity was observed in muscles spanning the stationary joint (shoulder muscles in experiment 1 and elbow muscles in experiment 2). This muscle activity preceded movement and varied in amplitude with the magnitude of upcoming interaction torque (the load resulting from motion of the nonstationary limb segment). In a third experiment, subjects performed multijoint movements involving simultaneous motion at the shoulder and elbow. Movement amplitude and velocity at one joint were held constant, while the direction of movement about the other joint was varied. When the direction of elbow motion was varied (flexion vs. extension) and shoulder kinematics were held constant, EMG activity in shoulder muscles varied depending on the direction of elbow motion (and hence the sign of the interaction torque arising at the shoulder). Similarly, EMG activity in elbow muscles varied depending on the direction of shoulder motion for movements in which elbow kinematics were held constant. The results from all three experiments support the idea that central control signals to muscles are adjusted, in a predictive manner, to compensate for interaction torques-loads arising at one joint that depend on motion about other joints.     

老年人左、右手喝水动作的运动特性研究

目的通过采集和分析人体三维运动数据,研究老年人左、右手喝水动作的运动学特性。分析左、右手动作的差异,为镜像康复设备的设计和精准康复训练方案的制定提供数据支持和理论基础。方法选取16名右利手的老年人为实验对象,实验对象分别用左、右手完成喝水动作。应用运动捕捉系统采集实验者上肢的三维运动轨迹,分析实验对象在喝水动作中,左、右肩、肘、腕关节的三维运动角度和角速度的相关性。结果左、右手在肩关节的矢状轴角度、肘关节的屈伸角度、肩关节的旋内旋外角度、肘关节旋内旋外的角度上具有高度相关性(相关系数r0.8);在肩关节冠状轴的角度、腕关节掌屈背伸的角速度上具有低度相关性(0.3r0.5);在肩关节冠状轴的角速度上具有微弱相关性(0.1r0.3);在其余上肢的关节角度和角速度上具有显著相关性(0.5r0.8)。结论健康老年人在完成喝水动作中,主要利用肩关节的旋内旋外活动和肘关节的屈伸、旋内旋外活动。右手在关节屈伸角度的运动幅度上大于左手,而在关节旋内旋外角度的运动幅度方面小于左手。在镜像康复机器人设计和康复轨迹规划中,应尊重左、右手的差异,实现精准康复的目的。     

Control of 3D limb dynamics in unconstrained overarm throws of different speeds performed by skilled baseball players

This study investigated how the human CNS organizes complex three-dimensional (3D) ball-throwing movements that require both speed and accuracy. Skilled baseball players threw a baseball to a target at three different speeds. Kinematic analysis revealed that the fingertip speed at ball release was mainly produced by trunk leftward rotation, shoulder internal rotation, elbow extension, and wrist flexion in all speed conditions. The study participants adjusted the angular velocities of these four motions to throw the balls at three different speeds. We also analyzed the dynamics of the 3D multijoint movements using a recently developed method called "nonorthogonal torque decomposition" that can clarify how angular acceleration about a joint coordinate axis (e.g., shoulder internal rotation) is generated by the muscle, gravity, and interaction torques. We found that the study participants utilized the interaction torque to generate larger angular velocities of the shoulder internal rotation, elbow extension, and wrist flexion. To increase the interaction torque acting at these joints, the ball throwers increased muscle torque at the shoulder and trunk but not at the elbow and wrist. These results indicates that skilled ball throwers adopted a hierarchical control in which the proximal muscle torques created a dynamic foundation for the entire limb motion and beneficial interaction torques for distal joint rotations.     

The leading joint hypothesis for spatial reaching arm motions

The leading joint hypothesis (LJH), developed for planar arm reaching, proposes that the interaction torques experienced by the proximal joint are low compared to the corresponding muscle torques. The human central nervous system could potentially ignore these interaction torques at the proximal (leading) joint with little effect on the wrist trajectory, simplifying joint-level control. This paper investigates the extension of the LJH to spatial reaching. In spatial motion, a number of terms in the governing equation (Euler’s angular momentum balance) that vanish for planar movements are non-trivial, so their contributions to the joint torque must be classified as net, interaction or muscle torque. This paper applies definitions from the literature to these torque components to establish a general classification for all terms in Euler’s equation. This classification is equally applicable to planar and spatial motion. Additionally, a rationale for excluding gravity torques from the torque analysis is provided. Subjects performed point-to-point reaching movements between targets whose locations ensured that the wrist paths lay in various portions of the arm’s spatial workspace. Movement kinematics were recorded using electromagnetic sensors located on the subject’s arm segments and thorax. The arm was modeled as a three-link kinematic chain with idealized spherical and revolute joints at the shoulder and elbow. Joint torque components were computed using inverse dynamics. Most movements were ‘shoulder-led’ in that the interaction torque impulse was significantly lower than the muscle torque impulse for the shoulder, but not the elbow. For the few elbow-led movements, the interaction impulse at the elbow was low, while that at the shoulder was high, and these typically involved large elbow and small shoulder displacements. These results support the LJH and extend it to spatial reaching motion.     

Cerebellar ataxia: torque deficiency or torque mismatch between joints?

Prior work has shown that cerebellar subjects have difficulty adjusting for interaction torques that occur during multi-jointed movements. The purpose of this study was to determine whether this deficit is due to a general inability to generate sufficient levels of phasic torque inability or due to an inability to generate muscle torques that predict and compensate for interaction torques. A second purpose was to determine whether reducing the number of moving joints by external mechanical fixation could improve cerebellar subjects' targeted limb movements. We studied control and cerebellar subjects making elbow flexion movements to touch a target under two conditions: 1) a shoulder free condition, which required only elbow flexion, although the shoulder joint was unconstrained and 2) a shoulder fixed condition, where the shoulder joint was mechanically stabilized so it could not move. We measured joint positions of the arm in the sagittal plane and electromyograms (EMGs) of shoulder and elbow muscles. Elbow and shoulder torques were estimated using inverse dynamics equations. In the shoulder free condition, cerebellar subjects made greater endpoint errors (primarily overshoots) than did controls. Cerebellar subjects' overshoot errors were largely due to unwanted flexion at the shoulder. The excessive shoulder flexion resulted from a torque mismatch, where larger shoulder muscle torques were produced at higher rates than would be appropriate for a given elbow movement. In the shoulder fixed condition, endpoint errors of cerebellar subjects and controls were comparable. The improved accuracy of cerebellar subjects was accompanied by reduced shoulder flexor muscle activity. Most of the correct cerebellar trials in the shoulder fixed condition were movements made using only muscles that flex the elbow. Our findings suggest that cerebellar subjects' poor shoulder control is due to an inability to generate muscle torques that predict and compensate for interaction torques, and not due to a general inability to generate sufficient levels of phasic torque. In addition, reducing the number of muscles to be controlled improved cerebellar ataxia.     

Multijoint movement control in Parkinson's disease

Impairments in the performance of complex actions in Parkinson's disease (PD) patients are well documented. The aim of the present study was to investigate potential mechanisms that may be contributing to impaired movement performance in PD patients. PD patients and age-matched control subjects performed rapid pointing movements to a series of four tabletop targets. The height of the table was adjusted until the targets could be achieved with arm movements in the horizontal plane. The targets were arranged such that target 1 required elbow extension only and targets 2–4 required increasing amounts of horizontal shoulder flexion in addition to the elbow extension. While the control subjects accelerated and decelerated the elbow and shoulder joints simultaneously regardless of the target location, the PD patients decomposed motion during the acceleration phase by accelerating first the shoulder and then the elbow joint. For PD patients this decomposition of arm segments was associated with greater coactivation of the muscles about the elbow when elbow extension and shoulder flexion were simultaneously required (targets 2–4), in contrast to the single joint action. The control subjects decreased elbow joint coactivation while the patients increased it across the four targets. The resulting peak interaction torques at both the elbow and shoulder joints occurred relatively later for the PD patients. The coactivation patterns observed in PD patients may reduce the ability to take advantage of interaction torques and may also contribute to joint motion decomposition. Electronic Publication     

Reprogramming of muscle activation patterns at the wrist in compensation for elbow reaction torques during planar two-joint arm movements

The relationship between wrist kinematics, dynamics and the pattern of muscle activation were examined during a two-joint planar movement in which the two joints moved in opposite directions, i.e. elbow flexion/wrist extension and elbow extension/wrist flexion. Elbow movements (ranging from 10 to 70 deg) and wrist movements (ranging from 10 to 50 deg) were performed during a visual, step-tracking task in which subjects were required to attend to the initial and final angles at each joint. As the elbow amplitude increased, wrist movement duration increased and the wrist movement trajectories became quite variable. Analysis of the torques acting at the wrist joint showed that elbow movements produced reaction torques acting in the same direction as the intended wrist movement. Distinct patterns of muscle activation were observed at the wrist joint that were dependent on the relative magnitude of the elbow reaction torque in relation to the net wrist torque. When the magnitude of the elbow reaction torque was quite small, the wrist agonist was activated first. As the magnitude of the elbow reaction torque increased, activity in the wrist agonist decreased significantly. In conditions where the elbow reaction torque was much larger than the net wrist torque, the wrist muscle torque reversed direction to oppose the intended movement. This reversal of wrist muscle torque was directly associated with a change in the pattern of muscle activation where the wrist antagonist was activated prior to the wrist agonist. Our findings indicate that motion of the elbow joint is an important consideration in planning wrist movement. Specifically, the selection of muscle activation patterns at the wrist is dependent on the relative magnitude and direction of the elbow reaction torque in relation to the direction of wrist motion.