Robotics and Autonomous Systems 58 (2010) 1223–1230 Contents lists available at ScienceDirect Robotics and Autonomous Systems journal homepage: www.elsevier.com/locate/robot A client–server architecture for remotely controlling a robot using a closed-loop system with a biological neuroprocessor Daniel de Santos c , Víctor Lorente b , Félix de la Paz c , José Manuel Cuadra c , José R. Álvarez-Sánchez c , Eduardo Fernández a , José M. Ferrández a,b,* a Instituto de Bioingeniería, Universidad Miguel Hernández, Elche, Spain b Dept. Electrónica y Tecnología de Computadores, Universidad Politécnica de Cartagena, Spain c Dept. Inteligencia Artificial, UNED, Spain article info Article history: Available online 16 September 2010 Keywords: Remote robotic control Neuroprocessor Cultured neural network Client–server architecture abstract This paper introduces an open-source real-time system that remotely controls a robot using human neuroblastoma cultures and a client–server architecture. Multielectrode array set-ups have been designed for direct culturing of neural cells over silicon or glass substrates, providing the capability simultaneously to stimulate and record populations of neural cells. However, it is very difficult to attach these neural cells to the robot structure due to the special conditions of the biological material. The main objective of this research is to build a client–server system for remotely connecting a robot to a neural culture in a closed-loop experimentation. The robot sensors will feed the biological neuroprocessor, while the neural activity will be used for guiding the robot, controlling in this way the robotic behaviour. © 2010 Elsevier B.V. All rights reserved. 1. Introduction Mammalian nervous systems exhibit complex computational functions including sensory functions, motor function and, in humans, abstract thought. In particular, pattern recognition ex- hibited in our olfactory, visual and auditory functions are of par- ticular interest to the electronic and computing communities. Meanwhile, several approaches are attempting to mimic/substitute sensory or neural elements (missing because of congenital state or due to pathological processes) in order to enable/restore functions by establishing neuro-electronic interfaces. Classical computational paradigms consist in serial and super- vised processing computations with high-frequency clock silicon processors, with moderate power consumption, and fixed circuit structure. In contrast, the brain uses millions of biological proces- sors, with dynamic structure, slow commutations compared with silicon circuits, low power consumption and unsupervised learn- ing. There have been numerous approaches to creating bioinspired parallel processing [1]. However, silicon provides a fundamentally different technological platform to that of neurobiology. Neurons – the core technology component – have a huge number of inter- connections compared to the three in traditional transistors. This provides considerably more computational power. Furthermore, * Corresponding author at: Instituto de Bioingeniería, Universidad Miguel Hernández, Elche, Spain. E-mail address: jm.ferrandez@upct.es (J.M. Ferrández). this extraordinary connectivity is coupled with natural unsuper- vised learning based on varying connective efficiency, and it has been used for robotic guidance [2–4]. This paper introduces an open-source real-time system that remotely controls a robot using human neuroblastoma cultures and a client–server architecture. Multielectrode array set-ups have been designed for direct culturing of neural cells over silicon or glass substrates, providing the capability simultaneously to stimulate and record populations of neural cells. However, it is very difficult to attach these neural cells to the robot structure due to the special conditions of the biological material. The main objective of this research is to build a client–server system for remotely connecting a robot to a neural culture in a closed- loop experimentation. The robot sensors will feed the biological neuroprocessor, while the neural activity will be used for guiding the robot, controlling in this way the robotic behaviour. 2. Learning in human neuroblastoma cultures 2.1. Experimental procedures The physiological function of neural cells is modulated by the underlying mechanisms of adaptation and reconfiguration in response to neural activity. Hebbian learning describes a basic mechanism for synaptic plasticity wherein an increase in synaptic efficacy arises from the presynaptic cell’s repeated and persistent stimulation of the postsynaptic cell. The theory is commonly 0921-8890/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.robot.2010.09.003