A small spiking neural network with LQR control applied to the acrobot
read more
Citations
Principles of Neural Science
Optimisation of a fuzzy logic controller using the Bees Algorithm
MBPOA-based LQR controller and its application to the double-parallel inverted pendulum system
Multi-objective design of state feedback controllers using reinforced quantum-behaved particle swarm optimization
Reinforcement Learning in Continuous State- and Action-Space
References
Numerical Recipes in C: The Art of Scientific Computing
Numerical Recipes in C: The Art of Scientific Computing
Principles of Neural Science
Principles of Neural Science
Related Papers (5)
Frequently Asked Questions (12)
Q2. How can a sensor neuron be encoded?
Assuming a constant input j to a sensor neuron, the spiking rate of the neuron can be calculated by finding the time betweenspikes.
Q3. What is the use of the leeky integrate-and-fire model for neurons?
The leaky integrate-and-fire model for neurons [8, 9] is used in their simulations since it is concise, simple to implement and fast to simulate.
Q4. How many spiking neurons have been used in the blue brain project?
networks of 10,000 spiking neurons have been used in the large-scale implementations of the blue brain project [16] to model cortical columns of the brain.
Q5. What is the potential of tunable delays?
In future studies tunable delays will also be used since they have the potential to increase the computational power of SNNs [14].
Q6. What is the area of acrobot swing-up control?
The area of acrobot swing-up control is quite advanced with many other existing successful techniques in addition to those mentioned previously, including sigmoid NN function approximation for reinforcement learning [4, 20], evolving a non-feedback vector of torque values [12], a fuzzy controller used to increase the acrobot’s energy [13], output zeroing based on angular momentum and rotation angle of center-of-mass [17].
Q7. how can a spiking neural network be used to bias learning towards the trajectory?
The trajectory presented can be used to devise a detailed fitness function and perhaps network architecture to bias learning towards the trajectory.
Q8. What is the encoding of analog input into spikes?
This form of encoding analog input into spikes is commonly known as rate encoding, where the firing rate of a sensor neuron is proportional to the input.
Q9. What is the acrobot’s position in the Fig. 5b plot?
In the bottom plot of Fig. 5b the authors can see that during the swing-up phase only extreme values of torque produced by the LQR are exploited.
Q10. How was the fitness of the acrobot determined?
The fitness for each individual chromosome was determined by transcribing it to the weights of the network andrunning a 20-s simulation (20,000 1-ms steps) of the acrobot under the control of the network.
Q11. Who has provided funding for this research?
Funding for this research has been supplied in part by the University of Newcastle Research Scholarship (UNRS) and by The ARC Centre for Complex Dynamic Systems and Control (CDSC).
Q12. How many parents did the fittest survive?
To elaborate, there was a population of 10 parents producing 70 offsprings each generation, and the top 10 fittest of the combination of parents and offspring survived to make up the parents for the next generation.