Our goal is to enable a robot to learn how to sequence its actions to perform high-level tasks specified as natural language instructions, given successful demonstrations from a human partner. Our novel neuro-symbolic solution GOALNET builds an iterative two-step approach that interleaves (i) inferring next subgoal predicate implied by the language instruction, for a given world state, and (ii) synthesizing a feasible subgoal-reaching plan from that state. The agent executes the plan, and the two steps are repeated. GOALNET combines (i) learning, where dense representations are acquired for language instruction and the world state via a neural network prediction model, enabling generalization to novel settings and (ii) planning, where the cause-effect modeling by a classical planner eschews irrelevant predicates, facilitating multi-stage decision making in large domains. GOALNET obtains 78% improvement in the goal reaching rate in comparison to several state-of-the-art approaches on benchmark data with multi-stage instructions. Further, GOALNET can generalize to novel instructions for scenes with unseen objects. Source code available at https://github. com/reail-iitd/goalnet.

A unique way evolutionary algorithms (EAs) are different from other search and optimization methods is their recombination operator. For real-parameter problems, it takes two or more high-performing population members and blends them to create one or more new solutions. Many real-parameter recombination operators have been proposed in the literature. Each operator involves at least a parameter that controls the extent of exploration (diversity) of the generated offspring population. It has been observed that different recombination operators and specific parameters produce the best performance for different problems. This fact imposes the user to use different operator and parameter combinations for every new problem. While an automated algorithm configuration method can be applied to find the best combination, in this paper, we propose an Ensembled Crossover based Evolutionary Algorithm (EnXEA), which considers a number of recombination operators simultaneously. Their parameter values and applies them with a probability updated adaptively in proportion to their success in creating better offspring solutions. Results on single-objective and multi-objective, constrained, and unconstrained problems indicate that EnXEA's performance is close to the best individual recombination operation for each problem. This alleviates the use of expensive parameter tuning either adaptively or manually for solving a new problem.