Josh C. Bongard

Resilient Machines Through Continuous Self-Modeling

Animals sustain the ability to operate after injury by creating qualitatively different compensatory behaviors. Although such robustness would be desirable in engineered systems, most machines fail in the face of unexpected damage. We describe a robot that can recover from such change autonomously, through continuous self-modeling. A four-legged machine uses actuation-sensation relationships to indirectly infer its own structure, and it then uses this self-model to generate forward locomotion. When a leg part is removed, it adapts the self-models, leading to the generation of alternative gaits. This concept may help develop more robust machines and shed light on self-modeling in animals.

Evolutionary robotics

Taking a biologically inspired approach to the design of autonomous, adaptive machines.

Automated reverse engineering of nonlinear dynamical systems

Complex nonlinear dynamics arise in many fields of science and engineering, but uncovering the underlying differential equations directly from observations poses a challenging task. The ability to symbolically model complex networked systems is key to understanding them, an open problem in many disciplines. Here we introduce for the first time a method that can automatically generate symbolic equations for a nonlinear coupled dynamical system directly from time series data. This method is applicable to any system that can be described using sets of ordinary nonlinear differential equations, and assumes that the (possibly noisy) time series of all variables are observable. Previous automated symbolic modeling approaches of coupled physical systems produced linear models or required a nonlinear model to be provided manually. The advance presented here is made possible by allowing the method to model each (possibly coupled) variable separately, intelligently perturbing and destabilizing the system to extract its less observable characteristics, and automatically simplifying the equations during modeling. We demonstrate this method on four simulated and two real systems spanning mechanics, ecology, and systems biology. Unlike numerical models, symbolic models have explanatory value, suggesting that automated “reverse engineering” approaches for model-free symbolic nonlinear system identification may play an increasing role in our ability to understand progressively more complex systems in the future.

Evolving modular genetic regulatory networks

We introduce a system that combines ontogenetic development and artificial evolution to automatically design robots in a physics-based, virtual environment. Through lesion experiments on the evolved agents, we demonstrate that the evolved genetic regulatory networks from successful evolutionary runs are more modular than those obtained from unsuccessful runs.

Three Dimensional Stochastic Reconfiguration of Modular Robots.

Here we introduce one simulated and two physical three-dimensional stochastic modular robot systems, all capable of self-assembly and self-reconfiguration. We assume that individual units can only draw power when attached to the growing structure, and have no means of actuation. Instead they are subject to random motion induced by the surrounding medium when unattached. We present a simulation environment with a flexible scripting language that allows for parallel and serial selfassembly and self-reconfiguration processes. We explore factors that govern the rate of assembly and reconfiguration, and show that self-reconfiguration can be exploited to accelerate the assembly of a particular shape, as compared with static self-assembly. We then demonstrate the ability of two different physical three-dimensional stochastic modular robot systems to self-reconfigure in a fluid. The second physical implementation is only composed of technologies that could be scaled down to achieve stochastic self-assembly and self-reconfiguration at the microscale.

Evolving Complete Agents using Artificial Ontogeny

In this report we introduce an artificial evolutionary system, Artificial Ontogeny (AO), that uses a developmental encoding scheme to translate a given genotype into a complete agent, which then acts in a physically-realistic virtual environment. Evolution is accomplished using a genetic algorithm, in which the genotypes are treated as genetic regulatory networks. The dynamics of the regulatory network direct the growth of the agent, and lead to the construction of both the morphology and neural control of the agent. We demonstrate that such a model can be used to evolve agents to perform non-trivial tasks, such as directed locomotion and block pushing in a noisy environment. It is shown that mutations expressed earlier in development tend to have a more variable morphological and behavioural effect than mutations expressed later in development, which tend to have a less pronounced effect. These results support the hypothesis that ontogeny provides artificial evolution with beneficial mutations that have varying degrees of phenotypic effect, depending on their onset of expression during development. In addition, we evolve agents using a fitness function which indirectly selects for increased size. In these agents we find evidence of functional specialization and repeated, differentiated structure. In the final section we argue that such a system would be a useful tool for the evolutionary design of morpo-functional machines.

Nonlinear system identification using coevolution of models and tests

We present a coevolutionary algorithm for inferring the topology and parameters of a wide range of hidden nonlinear systems with a minimum of experimentation on the target system. The algorithm synthesizes an explicit model directly from the observed data produced by intelligently generated tests. The algorithm is composed of two coevolving populations. One population evolves candidate models that estimate the structure of the hidden system. The second population evolves informative tests that either extract new information from the hidden system or elicit desirable behavior from it. The fitness of candidate models is their ability to explain behavior of the target system observed in response to all tests carried out so far; the fitness of candidate tests is their ability to make the models disagree in their predictions. We demonstrate the generality of this estimation-exploration algorithm by applying it to four different problems-grammar induction, gene network inference, evolutionary robotics, and robot damage recovery-and discuss how it overcomes several of the pathologies commonly found in other coevolutionary algorithms. We show that the algorithm is able to successfully infer and/or manipulate highly nonlinear hidden systems using very few tests, and that the benefit of this approach increases as the hidden systems possess more degrees of freedom, or become more biased or unobservable. The algorithm provides a systematic method for posing synthesis or analysis tasks to a coevolutionary system.

New Robotics: Design Principles for Intelligent Systems

New robotics is an approach to robotics that, in contrast to traditional robotics, employs ideas and principles from biology. While in the traditional approach there are generally accepted methods (e.g., from control theory), designing agents in the new robotics approach is still largely considered an art. In recent years, we have been developing a set of heuristics, or design principles, that on the one hand capture theoretical insights about intelligent (adaptive) behavior, and on the other provide guidance in actually designing and building systems. In this article we provide an overview of all the principles but focus on the principles of ecological balance, which concerns the relation between environment, morphology, materials, and control, and sensory-motor coordination, which concerns self-generated sensory stimulation as the agent interacts with the environment and which is a key to the development of high-level intelligence. As we argue, artificial evolution together with morphogenesis is not only nice to have but is in fact a necessary tool for designing embodied agents.

The road less travelled: morphology in the optimization of biped robot locomotion

Stable bipedal locomotion has been achieved using coupled evolution of morphology and control of a 5-link biped robot in a physics-based simulation environment. The robot was controlled by a closed loop recurrent neural network controller. The goal was to study the effect of macroscopic, midrange and microscopic changes in mass distribution along the biped skeleton to ascertain whether optimal morphology and control pairs could be discovered. The sensor-motor coupling determined that small changes in morphology manifest themselves as large changes in the performance of the biped, which were exploited by the optimization process. In this way, mechanical design and controller optimization were reduced to a single process, and more mutually optimized designs resulted. This work points to alternative routes for efficient automated and manual biped optimization.

Ensemble pruning via individual contribution ordering

An ensemble is a set of learned models that make decisions collectively. Although an ensemble is usually more accurate than a single learner, existing ensemble methods often tend to construct unnecessarily large ensembles, which increases the memory consumption and computational cost. Ensemble pruning tackles this problem by selecting a subset of ensemble members to form subensembles that are subject to less resource consumption and response time with accuracy that is similar to or better than the original ensemble. In this paper, we analyze the accuracy/diversity trade-off and prove that classifiers that are more accurate and make more predictions in the minority group are more important for subensemble construction. Based on the gained insights, a heuristic metric that considers both accuracy and diversity is proposed to explicitly evaluate each individual classifiers contribution to the whole ensemble. By incorporating ensemble members in decreasing order of their contributions, subensembles are formed such that users can select the top