Surviving in the real world

Citation

Loughlin, C. (2012), "Surviving in the real world", Industrial Robot, Vol. 39 No. 3. https://doi.org/10.1108/ir.2012.04939caa.001

Surviving in the real world

Article Type: Editorial From: Industrial Robot: An International Journal, Volume 39, Issue 3

The theme of this issue is “Mobile Robots” and we include three award winning papers from the 2011 CLAWAR conference (C2qlimbing and Walking Robots).

I am a great fan of simulation, and it has always been a major part of the majority of projects I have been involved with. For systems that include some form of sensory feedback I like to simulate all parts of the system, and have them sending data from one part of the system to another using the actual communication system, such as RS423 or Bluetooth, rather than sending data within software modules running on the one system.

Using this approach you can have a very high level of confidence that when one simulated part of the system is replaced with the real thing (i.e. the robot) that it has a good chance of not self-destructing. You can also simulate some sensor data, create a command for the robot and then have another simulation module work out what the original sensor data should have been. This is a robot engineer’s equivalent of an accountant’s double-entry book keeping, and it is an excellent way of debugging a system without risking expensive hardware.

However, robots, and mobile robots in particular, are all about operation in the real world, and there comes a time when we must put aside our simulations and boldly go where we have not gone before. Simulations can never replace actual trials in the real world and for this reason we have a journal policy of not publishing papers on work that has only been tested in simulation.

This will be considered by many to be a controversial and perhaps even unfair position, and I am well aware that we will be rejecting a lot of excellent work in the process.

It is an absolutely inherent part of any simulation that it will give you the answers you expect.

A simulation has various inputs and is programmed to produce various outputs based on the laws of physics, and takes into account all the factors that the simulation programmer thought should be applied. Some simulations can be highly sophisticated and incorporate real-world difficulties like wheel slip and thermal expansion – but they will always give you the answers you expect, and if they do not then it is a fault of the simulation that a few more hours coding will correct.

Moltke’s theory of war is famous for the observation that “no plan survives first contact with the enemy” and the same applies for simulations and real-world trials. The real world is all about the “unexpected” and so a simulation, which is all about the “expected” should only be considered a very useful stepping stone, and not an end result in itself.

I would therefore like to encourage all robotics researchers (and their supervisors) to make sure they test their ideas on actual prototypes. I fully understand that prototypes cost money and that research budgets are under stress at the moment, but it would be a great shame if a graduate engineer had never left the security of a software environment. Besides – the real world is much more fun.

Clive Loughlin