Electronic Skin, AI Rat and Teaching Pigs to Fly
December 12
This week we look at a new type of electronic skin that uses magnetic fields to allow detection of force on 3 axes leading to a new way for robots to learn manual skills. We examine a new AI driven robot that learnt how to interact with rats so much so that the rats accepted the robot as one of their own. We investigate a new method of desalinating water that allows greater production using the current infrastructure. Finally the silly season has begun in the lead up to Christmas so we will discover how OpenAI’s o1 large language model thinks it could teach a pig to fly. It gives us some insight into how the models are now thinking in a more structured way.
Electronic Skin
Electronic skin is designed to mimic the qualities of human skin in order to give robots new capabilities and to develop more advanced prosthetics. Flexible sensing materials that mimic our skin’s ability to pick up tactile information when touching objects and surfaces, create new possible applications in these areas.
At team at CNRS-University of Montpellier have developed a new soft e-skin that leverages magnetic fields to detect forces on three axes. The skins is made up of three layers. A top layer of flexible magnetic film, in the middle is an elastomer sheet and on the bottom is a printed circuit board.
When touching an object or surface the top magnetic field deforms. That deformation in turn affects the magnetic field beneath it and that allows the circuit board to register the changes due to the contact. This allows an estimation of the applied force. This design is more streamlined and easier to calibrate than other current designs.
The three dimensional measurement of forces could allow robots to learn new manual skills via touch based demonstrations. A robot may be taught to make coffee via a touch based demonstration.
In the future the system could be integrated with wearable technologies, prosthetics and robotic systems to improve their tactile sensing capabilities. This could open up a whole new range of robotic applications including humanoid robots working in spaces with humans.
AI driven Robot pretending to be a Rat
A team at the Beijing University of Technology in collaboration with a team at Technical University of Munich has created a rat robot that can interact with real rats in social ways. An AI model was used to train the robot to behave like a real rat.
The team used feedback loops combined with an AI reinforcement training to give the rat social skills that were strong enough to fool the rats into interacting with them. Research has shown that rats can be aggressive or playful and fights will break out in times of stress. Happy rats roll around on the floor wrestling with each other or nuzzling with their snouts. To fool the rats into believing that the robot is actually a rat, it would have to be able to mimic these behaviors convincingly.
The AI was trained on videos of rats doing what rats do when interacting. The robot rats learnt how to behave around other rats. The learning continued when the robot rat was exposed to real rats. The robot rat did not look exactly like other rats. It was cart like with wheels for feet. The spine could twist and turn like a rat and its’ forelimbs could interact physically as if it was a real rat.
In testing the rats accepted the robot rat and respond as they would be expected to respond. They would cower with fear when it appeared angry and wrestle and nestle just like rats would do with living cage mates during the calmer moments.
Science fiction shows robots that have the ability to interact with humans and make us forget that we are interacting with a robot. Today’s robots are nowhere near that level of capability. This may be a small first step towards that future.
Faster Seawater Desalination
A team from the University of South Australia and China have developed a way to desalinate seawater more quickly. Solar powered evaporation is an energy efficient alternative to the current desalination methods. One problem though is the slower evaporation rate for seawater compared to pure water. This is due to the negative effect of salt ions on water evaporation.
The team introduced an inexpensive and common clay mineral into a floating photo thermal hydrogel evaporator. This caused the evaporation rate of seawater to increase. They achieved evaporation rates for seawater 18.8% higher than for pure water. This compares to other evaporation systems where seawater evaporation was 8% slower than pure water.
The key is in the ion exchange process at the air-water interface. The minerals in the clay selectively enrich magnesium and calcium ions from the seawater at the evaporation surface which boosts the evaporation rate. The process occurs spontaneously which makes it highly convenient and cost effective.
There are about 17,000 operational desalination plants globally. Even small declines in desalination performance can affect output considerably. The new process can be integrated into existing infrastructure and will provide tens of millions of tons more clean water annually to clean water starved locations.
Teaching a Pig to Fly
As it is near the end of the year and we are all starting to wind down. I thought it might be fun to include something that I came across in The Neuron, a great daily AI newsletter. The authors asked OpenAI o1 (the latest and greatest version of their LLM) to describe the process involved in teaching a pig to fly. It gives you an idea of the state of the latest AI models and how they more systematically think through (even improbable) questions. Also it is a bit of fun. Please let me know how you go if you try this at home. Video proof preferred.
How to teach a Pig to fly (by OpenAI o1 via The Neuron)
Below is a step-by-step outline of a pragmatic, engineering-heavy approach. Although still highly speculative and ethically challenging, the focus will be on methods that rely on existing or near-future technologies, rather than waiting for extreme genetic engineering breakthroughs.
1. Establish a Controlled Research Environment
Rationale: Before attempting anything, you need the right setting.
Facility Setup:
A specialized hangar or large indoor test area with tethering systems, safety nets, and foam padding to prevent injury during initial tests.
Access to veterinary support, animal behavior experts, aerospace engineers, and mechanical/electrical engineers.
Bioethics board and regulatory compliance officers to ensure humane treatment and adherence to local laws.
2. Selection and Conditioning of the Subject Pig
Rationale: Start with a smaller, lighter pig for safety and feasibility.
Animal Selection:
Choose a small, healthy miniature pig breed (e.g., a “mini pig” weighing <20 kg) to reduce lift requirements.
Vet examination and baseline health metrics established: weight, muscle mass, cardiovascular fitness.
Physical Conditioning:
Gradually increase the pig’s activity level to improve cardiovascular health.
Use mild treadmill exercises and controlled diets to reduce excess body fat, aiming for a leaner, more athletic build.
Acclimate the pig to harnesses and gentle restraint to avoid stress when gear is introduced.
3. Design and Fabrication of the Prosthetic Flight Apparatus
Rationale: Engineering a flight solution that doesn’t rely on massive biological changes.
Wing Structure:
Develop lightweight carbon-fiber or advanced composite wings with a high aspect ratio to maximize lift.
Include adjustable wing surfaces (small flaps or ailerons) controlled by a servo system for fine-tuned maneuverability.
Harness and Exoskeleton:
A comfortable harness distributing load evenly across the pig’s torso, ideally shaped so the pig’s center of gravity aligns with the wing’s lift point.
Integrated mechanical support around shoulders and torso to anchor wing actuation systems without causing discomfort.
Propulsion and Control Systems:
Consider a hybrid approach: small electric ducted fans or micro-turbines for powered lift and forward thrust. These can be housed above or beside the pig’s body, integrated into the wing structure.
Energy Source: High-density lithium-sulfur or solid-state batteries, possibly with a tethered power supply for initial indoor tests.
Control Computer: A compact flight controller, similar to those in high-end drones, equipped with gyroscopes, accelerometers, and possibly lidar for altitude hold.
Wireless Remote Control: Initially, engineers on the ground operate the flight surfaces to stabilize and guide the pig during training flights.
4. Sensory and Safety Systems
Rationale: Ensure safety and minimize risk of injury.
Safety Tether:
Use a lightweight overhead cable that can support the pig in case of mechanical failure during early test flights.
Automatic descent mechanisms (small parachute or emergency auto-rotation system) if propulsion fails.
Comfort and Behavior Monitoring:
Integrate biometric sensors to monitor the pig’s heart rate, breathing, and stress levels.
Real-time telemetry so veterinarians and handlers can abort a flight if the pig becomes distressed.
5. Stepwise Training and Flight Testing
Rationale: Gradual acclimatization is essential.
Phase 1: Static Suspension:
Place the pig in the harness and wings without turning on any propulsion.
Let the pig get used to the feeling of being supported and partially suspended in the test environment.
Phase 2: Low-Level Assisted Lifts:
Use the propulsion system at minimal power to generate slight upward lift while still keeping the pig on a safety tether, barely leaving the ground.
Short sessions to avoid stress, gradually increasing lift and duration as the pig becomes comfortable.
Phase 3: Hover and Short Controlled Flights:
Attempt short hovers at a very low altitude (under 1 meter) to test stability and control system responsiveness.
Incrementally increase altitude and forward motion as control fidelity and the pig’s comfort improve.
6. Behavioral Reinforcement and Adaptation
Rationale: Pigs are intelligent and can adapt to routines.
Positive Reinforcement:
Use treats and familiar cues to help the pig associate the harness and mild lift with positive outcomes (food, gentle petting).
Over time, the pig may become less anxious and more tolerant of “flight” sessions.
Data Logging and Iterative Improvement:
Record all flight parameters: wing angles, thrust levels, pig’s biometrics.
Adjust wing shape, thrust vectoring, and harness fit based on data to optimize stability and reduce stress.
7. Scaling Up and Refinement
Rationale: Once basic flight is achieved, consider improvements.
Improved Efficiency:
Experiment with lighter materials, more efficient motors, and better battery technology.
Fine-tune aerodynamic surfaces for increased lift-to-drag ratio.
Autopilot and Assisted Flight Software:
Implement advanced stabilization algorithms so the pig’s flights are smoother, minimizing unexpected movements that could frighten it.
8. Ethical Considerations and Transparency
Rationale: Such a project is ethically sensitive.
Welfare Checks:
Continual veterinary oversight to ensure that the pig suffers no injuries, undue stress, or long-term health issues from the harness and test flights.
Public and Regulatory Compliance:
Obtain proper permits and comply with animal welfare regulations.
Be transparent about the purpose, methods, and outcomes with oversight committees and the public.
Summary
This approach hinges on starting small, using engineering solutions (prosthetic wings, powered lift, and careful flight control), and moving forward very slowly to ensure animal welfare. Instead of attempting biological transformations, the goal would be to augment the pig’s body with advanced materials and machinery that supply lift and thrust. Although this is a far cry from natural, unaided flight, it represents the most practical path to achieving something akin to porcine aerial locomotion with current or near-future technology.
Paying it Forward
If you have a start-up or know of a start-up that has a product ready for market please let me know. I would be happy to have a look and feature the startup in this newsletter. Also if any startups need introductions please get in touch and I will help where I can.
If you have any questions or comments please comment below.
I would also appreciate it if you could forward this newsletter to anyone that you think might be interested.
Till next week.