Evolutionary Search
A black-box optimization algorithm inspired by biological evolution that iteratively searches for optimal solutions through population selection, crossover, and mutation, and is commonly used in the design of robot morphologies and tools.
Timeline (last 90 days)
Some experiments use evolutionary search to optimize tool dimensions, such as a gripper's length, width, or joint angles.
Evolutionary search is a class of black-box optimization algorithms inspired by biological evolution, with typical representatives including genetic algorithms (GA), evolution strategies (ES), and CMA-ES.
The basic process of evolutionary search is: initialize a set of candidate solutions (a population), compute the fitness of each individual (here, task performance in simulation), then produce the next generation through selection, crossover/recombination, and mutation, converging to a better solution after multiple generations.
In agent framework optimization, distinguishing whether framework optimization succeeds more reliably within existing capabilities (elicitation) or expands the range of solvable tasks (expansion) is crucial for recursive self-improvement and evolutionary search.
All Facts (4)
Some experiments use evolutionary search to optimize tool dimensions, such as a gripper's length, width, or joint angles.
50%UnverifiedEvolutionary search is a class of black-box optimization algorithms inspired by biological evolution, with typical representatives including genetic algorithms (GA), evolution strategies (ES), and CMA-ES.
50%UnverifiedThe basic process of evolutionary search is: initialize a set of candidate solutions (a population), compute the fitness of each individual (here, task performance in simulation), then produce the next generation through selection, crossover/recombination, and mutation, converging to a better solution after multiple generations.
50%UnverifiedIn agent framework optimization, distinguishing whether framework optimization succeeds more reliably within existing capabilities (elicitation) or expands the range of solvable tasks (expansion) is crucial for recursive self-improvement and evolutionary search.
50%