How Real Transformers Work

The coolest thing about Transformers, of course, is that they can take two completely different shapes. Most can be bipedal robots or working vehicles. Some can instead transform into weapons or electronic devices. A Transformer's two forms have vastly different strengths and capabilities.

This is completely different from most real robots, which are usually only good at performing one task or a few related tasks. The Mars Exploration Rovers, for example, can do the following:

An Exploration Rover wouldn't be very good at tasks that don't fit into these categories. It can't, for example, assemble a bridge, fit into very small spaces or build other robots. In other words, it would make a lousy search-and-rescue robot, and it wouldn't fit in at all in an automated factory.

That's why engineers are developing reconfiguring robots. Like Transformers, these robots can change their shape to fit the task at hand. But instead of changing from one shape to one other shape, like a bipedal robot to a tractor-trailer, reconfiguring robots can take many shapes. They're much smaller than real Transformers would be; some reconfiguring robot modules are small enough to fit in a person's hand.

A module is essentially a small, relatively simple robot or piece of a robot. Modular robots are made of lots of these small, identical modules. A modular robot can consist of a few modules or many, depending on the robot's design and the task it needs to perform. Some modular robots currently exist only as computer simulations; others are still in the early stages of development. But they all operate on the same basic principle - lots of little robots can combine to create one big one.

Modules can't do much by themselves. A reconfiguring system also has to have:

Most modular, reconfiguring robots fit into one of three categories: chain, lattice and modular configuration. Chain robots are long chains that can connect to one another at specific points. Depending on the number of chains and where they connect, these robots can resemble snakes or spiders. They can also become rolling loops or bipedal, walking robots. A set of modular chains could navigate an obstacle course by crawling through a tunnel as a snake, crossing rocky terrain as a spider and riding a tricycle across a bridge as a biped.

Examples of chain robots are Palo Alto Research Center's (PARC) Polybot and Polypod and NASA's Snakebot. Most need a human or, in theory, another robot, to manually secure the connections with screws.

The basic idea of a lattice robot is that swarms of small, identical modules that can combine to form a larger robot. Several prototype lattice robots already exist, but some models exist only as computer simulations. Lattice robots move by crawling over one another, attaching to and detaching from connection points on neighboring robots. It's like the way the tiles move in a sliding tile puzzle. This method of movement is called substrate reconfiguration – the robots can move only along points within the lattice of robots. Lattice modules can either have self-contained power sources, or they can share power sources through their connections to other modules.

Lattice robots can move over difficult terrain by climbing over one another, following the shape of the terrain, or they can form a solid, stable surface to support other structures. Enough lattice robots can create just about any shape. Computer simulations show them changing from a pile of parts to a teacup and from a dog to a couch. The modules can combine to make flat surfaces, ladders, movable appendages and virtually any other imaginable shape. So a lattice robot is more like a Terminator T-1000 than a Transformer.

Robotics labs have created and theorized several lattice robot systems:

Like lattice robots, mobile reconfiguration robots are small, identical modules that can combine to form bigger robots. However, they don't need their neighbors' help to get from place to place – they can move around on their own. Mobile configuration robots are a lot like cartoon depictions of schools of fish or flocks of birds that combine to create a tool or structure. They move independently until they need to come together to accomplish a specific task. Swarm-bots, a project by the Future and Emerging Technologies program in the European Union, are mobile reconfiguration robots.

Even though these robots look very different from one another, they have many similarities in how they move and operate. We'll look at these next.

In addition to their size and modular structure, self-reconfiguring robots are different from Transformers in one major way. Optimus Prime and other Transformers are self-aware and can make independent decisions, and they keep their brains in one location within their bodies. A Transformer's brain controls each of its moving parts, and the parts themselves have little if any autonomy.

In most modular robot configurations, though, each module has some decision-making power and gets to help figure out where it's going to move. Instead of one module being the boss of all the others, planning and movement capabilities are distributed across all of the modules.

This idea -- a swarm of little robots, each of which gets to decide where it's going to go -- might sound disastrous. But the modules are programmed with a set of geometry-based rules about how to move. They're also programmed with algorithms that govern their movement. These algorithms and rules allow the robots to figure out how to change from one shape into another and to move across terrain.

For very complex maneuvers, the robots instead plan out a series of sub-shapes rather than trying to make a major change in one step. For example, a lattice robot that needs to change from a random pile of modules to a bipedal robot might first form the legs. Then, it might use those legs as a scaffold to build the upper half of the robot.

At this time, many of these robots can make simple transitions from one shape to another on their own. More complex changes might require the help of a scientist, making the collection of robots semi-autonomous rather than autonomous. A few robots that are still in the early stages of development receive all of their instruction from a computer workstation and make no decisions on their own.

Currently, most re-configuring robots have their own system of rules and algorithms, and rule sets work only for the robots for which they were designed. In other words, the rules for Rus Robotics Laboratory's Crystal won't work with Molecule.

However, scientists are using computer simulations to research movement theories that could work regardless of what a robotic module looks like. These theories could establish ground rules for robot movement, including:

If successful, this research could make it easier for engineers to make new, working modular robots that follow the same rules of movement.

Even though these robots don't walk and talk like Transformers do, they can change into virtually any shape, given the right programming and instructions. To find out more about them and about Transformers, see the links on the next page.