This week we examine a new type of electric engine that proceeds three times the power density of conventional electric engines. We also discover a new way of giving cars virtual x-ray vision and a new navigation system that will not rely on GPS. Finally we look at a new learning algorithm that allows legged robots to learn as they encounter new environments.
Tripling the Power of Electric Engines
H3X, a Denver based startup has developed a new electric motor that they claim has 3 times the power density of traditional electric motors.
Traditional electric motors can only generate a maximum of 4kw/kg (4 kilowatts per kilogram of weight). A commercial aircraft engine such as the engines on a Boeing 737 must deliver continuous power density of 12kw/kg. H3X’s HPDM-250 motor delivers a power density of 13kw/kg.
The team started from scratch to redesign an electric motor which usually comprise a gearbox, power delivery system and main motor. The components are housed separately to allow sufficient space for cooling thus avoiding engine failure. This makes the engines bulky and relatively heavy.
Using advances in material science and electronics the team 3D printed a variety of components in materials such as copper. All of the components were able to be housed in a single unit that weighs 15kg without reducing the cooling capacity.
The team envisages short term applications in Urban air mobility, electric boats and select regional aircraft markets. Longer term they aim to be able to power large commercial aircraft for short haul flights of up to 1,000 miles.
New Zealand Electric Plane Flight
Whilst we are talking about electric planes the first electric plane has flown from the North Island of New Zealand to the South Island. The 40 minute flight in a two seater aircraft came 101 years after the first conventional aircraft flight between the two islands. The plane flew at 100 feet above sea level at 130 kph (very slow for an aircraft) for the 78 kilometer trip. Wellington Airport is preparing for short hop electric plane flights to be a regular occurrence within the next 5 years.
X-Ray Vision for Cars
Cohda Wireless, the iMove Cooperative Research Center (part of Transport for NSW) and Sydney University’s Australian Center for Field Robotics have developed a new system for autonomous vehicles to see objects obscured behind other vehicles.
The system is based upon the collective perception (CP) of all the vehicles in the area. The vehicles share data on the movement of pedestrians, cyclists and other objects near each vehicle to give everyone a wider view of the area being navigated. This allows smart vehicles to overcome the physical and practical limitations of onboard sensors. Roadside sensors can give added depth to the information that the vehicles have available, particularly in busy pedestrian areas.
Bringing together the perception of multiple vehicles should lower the cost per vehicle of providing sufficient sensors to allow safe navigation for autonomous vehicles. Human driven vehicles may also be connected to enable additional perception without having to retrofit sensors.
GPS Free Navigation
Researchers at Sandia National Laboratories in the United States (New Mexico, California and elsewhere) have developed a new device that may allow for GPS free navigation.
The GPS system is owned an operated by the US Military. They can (but rarely do) turn it off at any time. As we have discussed before there are now alternative global systems (Europe, Russia and China, Japan and India are launching more local systems) however the GPS system is by far the dominant system in commercial use.
An avocado sized vacuum chamber contains a cloud of atoms that is used for navigational measurements. The device contains a number of quantum sensors, accelerometers and gyroscopes which are used to detect movement and thus allowing navigation.
Previous versions of this type of device used a powered vacuum pump to to whisk away molecules that leak in and wreck measurements. This makes it bulky and impractical. The physics of the new device takes place in a cubic centimetre. Anything larger is wasted space. Two small devices called getters, use chemical reactions to stop rouge molecules from entering the system. The getters can work without a power source.
The chamber itself is built out of titanium and sapphire. These materials are especially good at blocking out gasses like helium which can squeeze through stainless steel and pyrex glass. The team’s current goal is to keep the device sealed and operational for five years. This will be an important milestone in proving field readiness. During this period the streamlining of manufacturing will be important to ultimate success.
Robots that learn from their Environment
Legged robots have many advantages over other types of robots. They have the ability to travel long distances and navigate a wide range of land based environments. To date however the robots have only been trained to move in specific environments. Predicting all possible environmental conditions that a robot might encounter whist it is operating is very challenging.
Researchers at UC Berkeley have developed a reinforcement learning based computational technique that will allow legged robots to actively learn from their surrounding environment. This will allow them to continuously improve their locomotion skills.
This reinforcement learning approach builds upon a motion initiation framework previously developed at Berkeley. A new technique (the randomized ensembled double Q learning algorithm) from New York University has also been integrated. It is this algorithm that allows the robots to continuously learn from prior experience.
In the future this new technique could be used to significantly improve the locomotion skills of legged robots navigating over a range of terrains. This will be vital for the realization of the many visions for legged delivery robots that will eventually bring packages to our door.
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 email me via my website craigcarlyon.com or comment below.
I would also appreciate it if you could forward this newsletter to anyone that you think might be interested.
Till next week.