Modeling, Identification and Control, Vol. 40, No. 4, 2019, pp. 189–198, ISSN 1890–1328 Payload estimation using forcemyography sensors for control of upper-body exoskeleton in load carrying assistance Muhammad R. U. Islam 1 Shaoping Bai 1 1 Department of Materials and Production, Aalborg University, Aalborg, Denmark. E-mail: (mraza,shb)@mp.aau.dk Abstract In robotic assistive devices, the determination of required assistance is vital for proper functioning of assistive control. This paper presents a novel solution to measure conveniently and accurately carried payload in order to estimate the required assistance level. The payload is estimated using upper arm forcemyography (FMG) through a sensor band made of force sensitive resistors. The sensor band is worn on the upper arm and is able to measure the change of normal force applied due to muscle contraction. The readings of the sensor band are processed using support vector machine (SVM) regression technique to estimate the payload. The developed method was tested on human subjects, carrying a payload. Experiments were further conducted on an upper-body exoskeleton to provide the required assistance. The results show that the developed method is able to estimate the load carrying status, which can be used in exoskeleton control to provide effectively physical assistance needed. Keywords: Forcemyography, payload estimation, assistive exoskeleton, physical human-robot interaction. 1 Introduction With the advancement in robot technology, exoskele- tons are being developed for medical, industrial and service applications. Based on the applications, ex- oskeletons are categorized in three types i.e. rehabil- itation, assistance and power augmentation exoskele- tons Fan and Yin (2013); Hsieh et al. (2017); Cui et al. (2016); Keller et al. (2016); Huang et al. (2015); Castro et al. (2019); Christensen and Bai (2018); Gunasekara et al. (2012); Zhou et al. (2015). Rehabilitation and power augmentation exoskeletons are mainly focused on serving humans to regain their mobility and helping the users with extra power to enhance their capability, respectively Bai et al. (2018). In this work, our inter- est is to use exoskeletons to assist users, which can be either factory workers, elderly or person weak muscle strength, in load carrying tasks. For physical assistance exoskeletons, the determina- tion of required assistance level is one of primary con- cerns. In load carrying tasks, one method to determine required assistance level is by knowing the payload value and joint configuration. In existing upper body exoskeleton systems, payload information is acquired by integrating force sensors at the end-effector, where the weight is hanged on to the exoskeleton and not car- ried by the human Rosen et al. (2001); Lee et al. (2014). This method is useful in specific applications, partic- ularly heavy-load carrying tasks. The implementation of this method for daily routine activities or factory tasks is not feasible, where the user carries objects of different attributes. To overcome this challenge, Elec- tromyography (EMG) based estimation methods are used instead to determine joint torques and provide as- sistance through exoskeletons McDonald et al. (2017); Mangukiya et al. (2017); Leonardis et al. (2015); Ab- dallah et al. (2017); Mghames et al. (2017); Tang et al. (2014); Rahman et al. (2015); Li et al. (2013); Kiguchi doi:10.4173/mic.2019.4.1 c  2019 Norwegian Society of Automatic Control