Fig. 2: Showing how MLE can provide a diversity of robots for a diversity of environmental niches. | Nature Machine Intelligence

Fig. 2: Showing how MLE can provide a diversity of robots for a diversity of environmental niches.

From: Evolving embodied intelligence from materials to machines

Fig. 2

Christopher Michel (Antarctica); Jean-Baptiste Mouret (Amazon); and Dimitry B. (Sahara)

In this alternative MLE architecture, each level consists of only one heterogeneous library of solutions. Such architectures are likely to be possible in the further future, where boundaries between sensing, actuation and structure are collapsed to promote emergence and integration, at the cost of increased computational costs and combinatorial effort.