Representing high-dimensional data to intelligent prostheses and other wearable assistive robots: A first comparison of tile coding and selective Kanerva coding

Abstract

Prosthetic devices have advanced in their capabilities and in the number and type of sensors included in their design. As the space of sensorimotor data available to a conventional or machine learning prosthetic control system increases in dimensionality and complexity, it becomes increasingly important that this data be represented in a useful and computationally efficient way. Well structured sensory data allows prosthetic control systems to make informed, appropriate control decisions. In this study, we explore the impact that increased sensorimotor information has on current machine learning prosthetic control approaches. Specifically, we examine the effect that high-dimensional sensory data has on the computation time and prediction performance of a true-online temporal-difference learning prediction method as embedded within a resource-limited upper-limb prosthesis control system. We present results comparing tile coding, the dominant linear representation for real-time prosthetic machine learning, with a newly proposed modification to Kanerva coding that we call selective Kanerva coding. In addition to showing promising results for selective Kanerva coding, our results confirm potential limitations to tile coding as the number of sensory input dimensions increases. To our knowledge, this study is the first to explicitly examine representations for realtime machine learning prosthetic devices in general terms. This work therefore provides an important step towards forming an efficient prosthesis-eye view of the world, wherein prompt and accurate representations of high-dimensional data may be provided to machine learning control systems within artificial limbs and other assistive rehabilitation technologies.

Pavlovian Control of Intraspinal Microstimulation to Produce Over-Ground Walking

“…Pavlovian control are relatively recent advancements made in the field of computing science (Travnik and Pilarski 2017;Sutton et al 2011;van Seijen et al 2015;Modayil and Sutton 2014).…”

“…Selective Kanerva coding was chosen as it has proven to perform well online with a large number of sensors (Travnik and Pilarski 2017). It is also simple to implement and conceptualize.…”

“…In the field of rehabilitation, TD(λ) has been used to produce GVFs for upper-limb prostheses (Pilarski et al 2012;Pilarski et al 2013a;2013b;Sherstan and Pilarski 2014;Edwards et al 2016). TOTD offers an equivalence to the theoretical forward view of TD learning with negligible increase in computational cost (van Seijen et al 2015) and has been used to predict the shoulder angle of an upper-limb prosthesis (Travnik and Pilarski 2017). Pavlovian control has successively been used to control switching events of an upper-limb prosthesis in able-bodied study participants (Edwards et al 2013) and participants with an amputations (Edwards et al 2016).…”

“…First, the sensor values were normalized from their usable range to values between 0 and 1. Selective Kanerva coding was used to represent the normalized sensor values as a binary vector (Travnik and Pilarski 2017).…”

“…Kanerva coding is used, as was the case in this work, because it is not affected by the increase in dimensions that plagues traditional tile coding (Travnik and Pilarski 2017).…”