There is a new robot in town, and this is no ordinary robot. Not only that it is cute, but it also imbues lifelike behavior. These are called bots_alive. They are six-legged toy spiders that are remote-controlled, normally by a tiny infrared controller. The user will need to connect a marker to its crown, attach an IR blaster to the headphone port of the user’s smartphone, and then launch the app. From there, this is where the fun begins.
The app utilizes a vision system which tracks the robot’s location and that of the blocks also contained in the kit. The app also acts as the robot’s brain, as it dictates where it will go and how to get there. Bots alive is very simple, and so are the rules: the robot goes to blue blocks, and avoids the red ones. However, the behaviors and movements look as if it is much more complex and natural thanks to the frontloading of biomimetic patterns through other processes.
With this simplicity, however, can come an array of opportunities. Users can create little mazes and make the bots travel. Users can also allow it to follow a predetermined trail. Even more exciting is that if you happen to have a pair of “bots,” you can make them duel.
Currently, bots alive is still raising funds via Kickstarter. Bots_alive is supported by a National Science Foundation SBIR grant. The bots_alive kit converts Hexbug Spiders into smart toys. These robots can navigate their environment with the help of the kit. What makes these robots specials is that they can be autonomous and independent. These hex bug spiders can scamper around and make their decisions. Apart from that, they can also learn through interactions. They can change its personalities and capabilities.
The artificial intelligence found on these robots is created in a unique a novel way. Knox writes in Robohub, Although these can be independent, there is a human teleoperator for the robots. The teleoperator helps navigate through the different scenarios. The teleoperator does not drive the robots, but follows certain heuristics algorithms and responds to events. However, there is some difference with Knox’s creation, which he underlines, The teleoperator of the bots_alive watches a screen displaying what the robots see. He can send commands by pushing buttons.
Through these teleoperation sessions, training data is gathered. These contain teleoperation commands; and information about reasons for giving the command. Machine learning is applied by the bots to create an internal understanding of the commands and reasons for the commands. This creates a heuristic thinking which answers the causal question of the environment: Within a given context, what are the chances that the teleoperator would give a certain command?
The robots are based on the behaviors of humans and animals. Knox and his team had the computer vision system observe as a human steered a “bot” through several scenarios involving blue and red blocks.
The bots_alive kit Kickstarter may be funded for as low as $35. For $60 you can get a Hexbug included in the box. After the crowdfunding, people will be able to follow and order at the bots_alive site.
These are no ordinary robots. These are also not ordinary AI. In one sense, these are hobbled AI because they are not capable of storing a large amount of information which can be used to determine future action. The AI is dependent on the rules and behavior of the commands from the smartphone.
In the same manner, the robots are programmable but only via the use of the AI. If you understand the behavior of the AI, then you can program the robots to do as you wish. The whole point is that these are not straightforward commands. Rather, these are standard reactions to a set of rules. Commands are then given which are also based on the set of rules.
This behavior-based programming or behavior-based AI is better as a learning tool for both robotics, and for AI. The user can learn a lot about programming using a learning system. In the same manner, understanding the underlying logic of how the robot thinks allows the user to do more things with fewer commands. The behavior is based less on the commands and more about how it will work out a problem when faced with different conditions. It obeys the commands by “thinking” on its own.
This is a big leap forward for young children to grasp. However, it also teaches them about AI in a deeper and much better context. Children will be able to create mazes, instructional commands, as well as trails and rules for the bots. This is one way they will be able to study programming, as well as to have first steps in robotics. Although simple looking in concepts, these also serve as the foundation tools for further studies. With more than one bots_alive kit, the interaction can also be studied, as well as hive behavior. Children’s creativity can be fired up with these toys. Hopefully, this can lead them to a better understanding of technology, computers, programming as well as robotics.
The use of the HexBug as a teaching tool also has a surprising side-effect. With only a small set of instructions, the user can let his imagination free and run the HexBugs across the bots_alive landscape. With the smartphone remote controller, the user can experiment with different combinations of commands and see where this would go.
