Bionic Cyber Physical Systems

I have been neglecting my blog because of several paper and proposal deadlines. Rather than keeping the blog gather more dust, I decided to share some summaries from recent talks I attended and papers I read. This is a talk I attended couple months ago and enjoyed a lot. It is directly on the distributed systems topic, but in the future we may find ourselves programming these kind of distributed systems, and we will be dealing with a lot of bugs in our code for sure ;-)

Alper Bozkurt (NCSU) has been working on bionic cyber physical systems, which aim to fuse synthetic man-made systems with naturally occuring biological organisms, such as cockroaches, moths, dogs, lemurs, and eventually humans. His research is very interesting and has been featured at National Geographics, CNN, and Stephen Hawkins show "Brave New World". Here is the PhDComics coverage of his research (and this).

Alper primarily works with insects. He says insects are fabuluous because they are optimized aerodynamic systems with onboard efficient control systems and dense chemical fat stores. In other words, insects have sensors and actuators, collision avoidance, and power source. Any plane-like machine at the size of insect will not fly because of aerodynamics. At that size, you need flapping wings or helicopter blades.

He makes a good case for converting the insects to biobots. We, humans, have used ATP powered animal muscle before oil powered engines through the use domestication technology. We domesticated horses, donkeys, mules, dogs. But we couldn't domesticate the insects, because they are hard to train and teach. He argues that the biobots technology is just a domestication technology (a computer-assisted technology) applied on insects. Domesticating insects via biobot technology has applications in search and rescue, environmental sensing, industrial plant monitoring, and nuclear plant monitoring. Think of a colony of biobot insects networked together for exploration and mapping for challenging terrains. The cost of a raising insects is almost nothing. Starting with a male and female insect, you can go to a colony in two weeks. Alper's research is on developing low cost neurostimulation probes that can be easily applied to the insects. The vision is that a robot would be able to catch the insects and implant these probes to the insects with a minimally invasive way. Insects are simplistic, they only respond to reflexes, so the neuro control is much simpler to perform in insects. Insects also don't register pain, there are no pain receptors in insects, so there is no need for permissions for running scientific experiments on insects.

The payload that can be carried by the insect is not much, but now thanks to miniaturization in circuit technology with a few grams of payload, it is possible to store and carry megabytes of digital information (photos/videos). The challenge is how to implant the circuits to the insects without crippling or killing them in the process. Alper's research solves this problem by benefiting from the metamorphic development of the insect. During metamorphosis, 90% of the tissue is retained and in 1 week the terrastrial insect morphs to become a flying insect by developing wings and antenna. To explain how drastic metamorphosis, people use the analogy of transforming from wheelchair to an Apache helicopter. Alper's method performs the surgery before metamorphosis, and the wound caused for surgery is auto-recovered. The surgery does not require skill, and it will possible to automate the surgery to be performed by a robot.

They work with Carolina sphinx moth, which is 4-7 cm, with a wing spand of 10cm. This is quite a large insect and weighs 1-2 grams, and can carry a payload capacity half of their weight. They fly at 5m/s speed. With the probe installed, Alper's team can control the moth to initiate flight, stop flight, turn right, left.

For terrestrial locomotion control, the team works with pet cockroaches, called hissing cockroaches. These cockroaches have 5-7 cm size and weigh 5-10 grams. The insect looks down (doesn't need to look for predators) uses antennas for sensing like blind people with canes. If the probe gives the right antenna the signal, the insect turns to left. Thus, the team is able to make the insect follow the line very precisely controlled by a joystick.

The tissue electrode interface is a bottleneck in the implants. The team is experimenting with electroplating to optimize enough charge and charge storage. They also run some in-vivo electrochemical analysis to see if they can improve/optimize the process. Finally, habituation can become a problem: the insect may get used to the implant, and may start to ignore after a while. The control is not always very deterministic, so they try to automate and make the control system adaptive.

The team is also developing backpack implants for the insects with zigbee (and sometimes Bluetooth support). Networking the insects will enable controlling them as a swarm and move them in a coordinated manner.


Popular posts from this blog

Graviton2 and Graviton3

Foundational distributed systems papers

Learning a technical subject

Learning about distributed systems: where to start?

Strict-serializability, but at what cost, for what purpose?

CockroachDB: The Resilient Geo-Distributed SQL Database

Amazon Aurora: Design Considerations + On Avoiding Distributed Consensus for I/Os, Commits, and Membership Changes

Warp: Lightweight Multi-Key Transactions for Key-Value Stores

Anna: A Key-Value Store For Any Scale

Your attitude determines your success