Nah, I thought this has got to be a joke:
The Pentagon’s defence scientists want to create an army of cyber-insects that can be remotely controlled to check out explosives and send transmissions.
But no…there is actually a DARPA call for proposals.
DARPA seeks innovative proposals to develop technology to create insect-cyborgs, possibly enabled by intimately integrating microsystems within insects, during their early stages of metamorphoses. The healing processes from one metamorphic stage to the next stage are expected to yield more reliable bio-electromechanical interface to insects, as compared to adhesively bonded systems to adult insects. Once these platforms are integrated, various microsystem payloads can be mounted on the platforms with the goal of controlling insect locomotion, sense local environment, and scavenge power. Multidisciplinary teams of engineers, physicists, and biologists are expected to work together to develop new technologies utilizing insect biology, while developing foundations for the new field of insect cyborg engineering. The HI-MEMS may also serve as vehicles to conduct research to answer basic questions in biology.
The final demonstration goal of the HI-MEMS program is the delivery of an insect within five meters of a specific target located at hundred meters away, using electronic remote control, and/or global positioning system (GPS). Although flying insects are of great interest (e.g. moths and dragonflies), hopping and swimming insects could also meet final demonstration goals. In conjunction with delivery, the insect must remain stationary either indefinitely or until otherwise instructed. The insect-cyborg must also be able to transmit data from DOD relevant sensors, yielding information about the local environment. These sensors can include gas sensors, microphones, video, etc.
Although the idea of having a remote controlled dragonfly is very cool, I am very pessimistic, and have to dash a little cold water on the plan.
OK, so the idea is to put a small chip into a larva, and when it undergoes metamorphosis and reorganizes its nervous system, it would integrate itself with the circuit in such a way that signals transmitted to the chip could control its behavior. I can sort of see what they want to do, but no—it’s pure fantasy. We don’t have the understanding of insect neurodevelopment to be able to even come close to what they want.
Chips aren’t going to be integrated, they’re going to be encapsulated. I’ve done long-term studies of larval insect nervous systems, and they’ve got a surprisingly good immune system—hemocytes are going to flock to whatever is implanted, and it’s going to be well glopped up with cells and proteins. It’s not an absolutely insurmountable problem, but it’s going to take some fancy materials science work and some kind of biologically-relevant coating to make it work.
To get any kind of reliable connection with neuronal elements, this chip is going to have to be layered with appropriate signaling/pathfinding molecules. I suppose that could be done, but you’d have to know what molecules would work, and you’d have to be able to pattern a chip with them. We know some of the molecules involved, but most of the work done with them have been disruptions, not attempts to guide axons down novel paths. It’s been done with neurons in vitro, but seems impractical in vivo. Once upon a time, I teased axons into growing down aberrant pathways, but that was all done with microsurgery, not molecules. Again, maybe someday something limited could be done along these lines, but it’s a long way off.
If we could string subsets of axons along to form connections with the circuitry on a chip, we still wouldn’t know how to control the animal. Even an insect nervous system is incredibly complex, full of utterly mysterious dendrites and axons and synapses that do important things, we just don’t know what. This is like sending some guy who knows next to nothing about avionics into a 747 with a pickaxe, a voltmeter, and a 9V battery, and telling him to hack into the wiring and take control of the plane. It may not be impossible, but it is the next best thing.
Given our man with a pickaxe, the other problem is that even if he does find the right wire to control the left aileron, he’s probably going to make such a hash of everything else on the way to finding it that he’s going to render the plane unflyable. Same with this insect; if the problem of incorporating a foreign body is overcome, if the problem of gently and biologically extracting a useful subset of the insect nervous system is overcome, if the problem of understanding how to impose a new and useful signal on that CNS is overcome, you’re still left with a bug whose brain and ganglia have been partially scrambled. It’s not likely to be viable.
It’s a ridiculously pie-in-the-sky idea, and someone in the Pentagon has clearly abandoned the reality-based community to come up with this one. There are little bits and pieces of the proposal that would be interesting to pursue, but as a whole, it’s weirdly impractical nonsense. The guy behind it all seems to be competent, but what he does for real doesn’t seem to apply very well to what these proposals demand—he’s working with chips that are far larger and cruder than what this proposal would require.
It’s a very weird proposal, and antithetical to my experience with government grants. Usually what’s wanted is something that is extraordinarily well-justified, that is demonstrably doable, that advances from a solid, well-established foundation…and this one expects the opposite:
Proposed research should investigate innovative approaches that enable revolutionary advances in science, devices or systems. Specifically excluded is research, which primarily results in evolutionary improvement upon existing state-of-the-art.
That sounds like a call for crackpots to me.