Should newborn robots inherit what their parents learned?
Letting simulated robots inherit the gait controllers their parents had learned made offspring walk faster than starting from inherited-but-unlearned controllers, especially when learning time was short.
Source
Lamarckian Evolution of Simulated Modular Robots
Study at a glance
- Design
- Computational / modelling — Simulated evolution of modular robot bodies and CPG/CPPN controllers in Revolve/Gazebo; Darwinian vs Lamarckian controller inheritance compared over five lineages and three HyperNEAT learning budgets.
- N
- Five lineages per condition, each with 20 robots per generation for 10 generations (200 robots per lineage); no single analytic N.
- Population
- Simulated modular robots built from RoboGen-style components
- Outcome
- Locomotion speed (fitness) after lifetime learning; parent-offspring morphological similarity; retention of parental controller material
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
In simulation, robot bodies built from bricks and motorised hinges were evolved for walking speed, and every new robot had a short learning period (using the HyperNEAT algorithm) to tune its controller. The controller combined a body-specific oscillator network with a body-independent network that could be passed between different bodies. In the Darwinian version, offspring received the controllers their parents were born with; in the Lamarckian version, they received the controllers their parents had learned. Both versions were run on the same five starting populations with small, medium and large learning budgets.
What they found
Lamarckian inheritance produced faster robots across generations, with the biggest advantage under small and medium learning budgets; with the largest budget the Darwinian system mostly caught up. The benefit of inheriting a learned controller was larger when a child's body was more similar to its parents'. Lamarckian runs also converged on more consistent body shapes, and offspring kept more of their parents' controller structure. A few body types still learned better from scratch.
The limits
What it doesn't show
Everything happens in one simulator with one task (straight-line speed), open-loop controllers without sensory feedback, and small populations of 20 over 10 generations, so results may not transfer to real hardware or other tasks; the authors say results are based on one system only. Comparisons rest mainly on plotted trends across only five lineages rather than formal significance tests of the main performance difference. The text supplied here lacks the paper's introduction and related-work sections.
Key terms
- Lamarckian evolution
- Inheritance in which traits acquired during an individual's lifetime, here learned controllers, are passed on to offspring.
- Darwinian evolution
- Inheritance in which offspring receive only the traits their parents were born with, not what they learned.
- Central pattern generator (CPG)
- A network of coupled oscillators that produces rhythmic motor output, used here to drive each robot joint.
- CPPN / HyperNEAT
- A compositional pattern-producing network that outputs controller weights from node coordinates; HyperNEAT evolves such networks.
- Learning budget
- The number of controller evaluations a newborn robot is allowed during its learning period.
Flashcards
0 of 9 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
In the Lamarckian setup, what does a newborn robot receive?
Common questions
Why can controllers be passed between robots with different bodies?
The CPPN part takes joint coordinates as input and outputs weights, so it does not depend on a fixed body layout; any CPPN can be applied to any body's oscillator network.
Is Lamarckian inheritance always better?
Not always. Its edge shrank with a large learning budget, and some bodies learned faster from scratch, especially when they differed a lot from their parents.
Why does this matter for real robots?
Physical trials are slow and costly, so reducing how many trials a new robot needs to learn to walk is valuable; inheriting learned controllers is one way to cut that cost.
More on Robot learning