Insect inspired navigation strategies have the potential to unlock robotic navigation in power-constrained scenarios as they can function effectively with limited computational resources. One such strategy, familiarity-based navigation, has successfully navigated routes of up to 60m using a single layer neural network trained with an Infomax learning rule in online robotic applications. Here we challenge Infomax to navigate longer routes, investigating the relationship between performance, view size, view acquisition rate and network size. By doing so, we determine the parameters at which Infomax operates effectively and explore the profile with which it fails. We show that effective memorisation of familiar views is possible for longer routes than previously attempted, but that this length decreases for reduced input view dimensions. In the selection of an ideal view acquisition rate, we also show that this must be increased with route length for consistent performance. In investigating the applicability to small, lower-power robots, we demonstrate that computational and memory savings may be made with equivalent performance by reducing the network size. Finally, we investigate the profile with which failure occurs, demonstrating increased confusion occurring across the route as it extends in length. These findings are being used to inform theories of insect navigation and improve practical deployment of view based navigation for long routes.

Having previously developed and tested insect-inspired visual navigation algorithms for ground-based agents, we here investigate their robustness when applied to agents moving in three dimensions, to assess if they are applicable to both flying insects and robots, focusing on the impact and potential utility of changes in height. We first demonstrate that a robot implementing a route navigation algorithm can successfully navigate a route through an indoor environment at a variety of heights, even using images saved at different heights. We show that that in our environments, the efficacy of route navigation is increased with increasing height and also, for those environments, that there is better transfer of information when using images learnt at a high height to navigate when flying lower, than the other way around. This suggests that there is perhaps an adaptive value to the storing and use of views from increased height. To assess the limits to this result, we show that it is possible for a ground-based robot to recover the correct heading when using goal images stored from the perspective of a quadcopter. Through the robustness of this bio-inspired algorithm, we thus demonstrate the benefits of the ALife approach.

Time varying artificial neural networks are commonly used for dynamic problems such as games controllers and robotics as they give the controller a memory of what occurred in previous states which is important as actions in previous states can influence the final success of the agent. Because of this temporal dependence, methods such as back-propagation can be difficult to use to optimise network parameters and so genetic algorithms (GAs) are often used instead. While recurrent neural networks (RNNs) are a common network used with GAs, long short term memory (LSTM) networks have had less attention. Since, LSTM networks have a wide range of temporal dynamics, in this paper, we evolve an LSTM network as a controller for a lunar lander task with two evolutionary algorithms: a steady state GA (SSGA) and an evolutionary strategy (ES). Due to the presence of a large local optima in the fitness space, we implemented an incremental fitness scheme to both evolutionary algorithms. We also compare the behaviour and evolutionary progress of the LSTM with the behaviour of an RNN evolved via NEAT and ES with the same fitness function. LSTMs proved themselves to be evolvable on such tasks, though the SSGA solution was outperformed by the RNN. However, despite using an incremental scheme, the ES developed solutions far better than both showing that ES can be used both for incremental fitness and for LSTMs and RNNs on dynamic tasks.

Insect-Inspired models of visual navigation, that operate by scanning for familiar views of the world, have been shown to be capable of robust route navigation in simulation. These familiarity-based navigation algorithms operate by training an artificial neural network (ANN) with views from a training route, so that it can then output a familiarity score for any new view. In this paper we show that such an algorithm – with all computation performed on a small low-power robot – is capable of delivering reliable direction information along real-world outdoor routes, even when scenes contain few local landmarks and have high-levels of noise (from variable lighting and terrain). Indeed, routes can be precisely recapitulated and we show that the required computation and storage does not increase with the number of training views. Thus the ANN provides a compact representation of the knowledge needed to traverse a route. In fact, rather than losing information, there are instances where the use of an ANN ameliorates the problems of sub optimal paths caused by tortuous training routes. Our results suggest the feasibility of familiarity-based navigation for long-range autonomous visual homing.