The innate intelligence of ants is helping Australian-based scientists develop prosthetic limbs that respond to brain signals in groundbreaking research that could change the lives of amputees.
The technology, created by a team of five researchers from the University of Technology Sydney (UTS), mimics the myoelectric signals used by the central nervous system (CNS) to control muscle activity.
Complex algorithms model the so-called "swarm-intelligence" used by ant colonies to locate food. Artificial intelligence researchers have long used the complex interactions between ants to construct a pattern recognition formula to identify bioelectric signals, which can then be applied in live human trials.
I don't think the crossover from science fiction to science reality is that far away now
PhD student Rami Khushaba, from the UTS Faculty of Engineering and Information Technology, said the behaviour of social insects, like ants, allows scientists to understand the body's electrical signals and use the knowledge to create a robotic prosthetic device that can be operated by human thought, like a flesh and blood limb.
"I don't think the crossover from science fiction to science reality is that far away now," Khushaba said.
Khushaba is developing the mathematical basis for identifying what biosignals relate to particular arm movements and where electrodes should be placed. Swarm intelligence algorithms were chosen for their abundance in nature and because they use multi-agent techniques to solve a specific problem.
“We can use the behaviour of the ants to enhance the quality of the control systems that we employ with the robotic limbs. The biggest problem in our field is the amputee acceptance rate — they are disappointed if the system is not fast and accurate enough,” Khushaba said.
The researchers create a map of the voluntary intent of the CNS called an electromyogram (EMG) by attaching sensors to the human forearm — or what remains of it after an amputation.
Khushaba said the technology uses wave-length transforms to extract the valuable information from captured data.
“We do many pre-processing techniques, we filter them, clean the data, and start to extract the important variables,” Khushaba said.
A simple microprocessor is mounted inside the hardware within the artificial limb – in this instance a forearm. While Khushaba would not be drawn on the cost of the system, he said it will not be expensive. He said the prosthetic limb will be available within two years once a manufacturer is found.
The team collected data on 10 movements using 10 variables of the forearm from six subjects and achieved a 99.9 percent accuracy. Khushaba said the movement of the forearm is captured and filtered against the variables to minimise processing time.
Only a few seconds of data is required to capture and train the system to identify patterns in the raw data during the online testing phase. The entire system operates on the Matlab programming language designed for technical computing.
“You get signals from each censor mounted on the forearm, and we describe these by variables. Then we have an objective function to get the best accuracy and we select the minimum number of these variables that will give us the highest classification accuracy,” Khushaba said.
He said the biggest challenge to the success of the robotic limb is maintaining system accuracy and speed. The EMG system extracts the signals from human muscles in less than 256 milliseconds.
"We hope accuracy will improve. It will be the very near future when amputees, who can still imagine moving a lost limb, will have access to a device that can truly respond to their intentions."