Thinking About Future Posts

Two areas that I'm going to be researching for future posts will be locomotion and robot muscle; specifically those types of each that will be adaptable to the needs of a robot harvester/tender.

Regarding modes of locomotion, I've already concluded that a walker will be the best all-around option for a harvester/tender [1],[2].  I've been looking at both hexapod walkers and robot lizard walkers and comparing their motion to that of actual lizards.  Three things stand out immediately.  First, real lizards have feet; hexapod walkers don't. Second, a real lizard's elbow moves horizontally, while the elbows on hexapod walkers move vertically.  And third, hexapod walkers have rigid torsos while a real lizard bends its spine from side to side as it moves.  What does all this mean?  I don't know.  Hopefully, with some research, thought, and time, I can arrive at some answers.

Once the mode of locomotion for a robot harvester/tender has been decided upon, the next step in design is to specify the type of actuator (robot muscle) that can be used to implement this motion.

Regarding possible candidates for robot muscle, to date there are no good candidates!  But the first step in narrowing down the set of most likely candidates is to start a list of the constraints that a viable candidate for a robot harvester/tender’s muscle must satisfy. 

Next, the question of our robot harvester/tender’s brain needs to be addressed.  Whatever computer system ends up being used to control a robot harvester/tender, it will still be held to the same constraints, in terms of size, weight, power consumption, cost, and etc., that the rest of the robot design will be held to.  But here lies the design challenge; there are no currently available computer systems that fit these constraints! 

Lastly, the question of fabrication methods: if/when robot harvester/tenders take their place in the farming industry, they will need to be produced by the 100K’s quantities.  They will need to be fabricated in a way that makes them cheap to make but still robust in use.  The modern auto industry has been addressing this exact issue for years, so a good place to start would be to look at how cars are designed and built.

[1] Agricultural Robots: Locomotion, Walker, Crawler or Wheels?
[2] Dancing Robots vs Robot Dancing

Piezoelectric Actuators

Electromagnetic driven actuators will be off the table for agricultural robotics because of their high current draw and low force/current ratios.  On the other hand, because of their high force/current and high force/size ratios, piezoelectric driven hydraulics has the potential to be a good candidate for "robot muscle" for portable and field deployed robotic systems.  The other nice thing about hydraulics is they are very abuse tolerant, work in the mud and dirt, and even under water. 

This is going to be an ongoing area of research for me.  As I find new information, I'll post it here and then bump this post to the head of the queue again.

This isn't hydraulics, but it is an interesting use of piezoelectrics to drive a linear actuator.  It uses a ‘walking” motion to generate 8-Nt of force with a 20-cm/sec velocity. http://www.linear-actuator.net/Ultrasonic-Motors_Linear_Actuators.php

Here is a piezoelectric driven hydraulic pump capable of 2500-psi output pressure and a flow rate of 40-cc/sec: http://www.kineticceramics.com/piez_fluid.html

Also see: http://aml.seas.ucla.edu/research/areas/piezoelectric/pump-non-mech-valves/2002-12_lee.pdf

Micro Gas Turbine Engines

While researching gas turbines, I came across a fascinating new area in high-tech R&D that bills itself as "micro turbines".  These are gas turbines on the order of 1 cm in size and turn at 500,000 to 1,000,000 RPM and higher.  Here are a few images from the links included below.

 

This combustion chamber only measures 20 mm in diameter and is 2.7 mm thick.
Supplied by a hydrogen-air mixture, it produces a power of up to 1200 W.



How about a 50-100 W gas turbine generator with

fuel tank the size of a pack of playing cards!

Here is a list of micro turbine links.  Some are news posts, but most are journal published, research papers.

Engine on a Chip

Micro-turbines, Maxi-difficulties

PowerMEMS Project - Microsystems for power generation

ULTRA MICRO GAS TURBINE GENERATOR

A SILICON-BASED MICRO GAS TURBINE ENGINE FOR POWER GENERATION (PDF)

 

DEVELOPMENT OF MICROMACHINE GAS TURBINES AT TOHOKU UNIVERSITY (PDF)

MILLIMETER-SCALE, MEMS GAS TURBINE ENGINES (PDF)

POWER MEMS AND MICROENGINES (PDF)

THE USE OF MICRO TURBINE GENERATORS IN HYBRID ELECTRIC VEHICLES (PDF)

DEVELOPMENT OF MICROMACHINE GAS TURBINE FOR PORTABLE POWER GENERATION (PDF)

ULTRA-HIGH-ENERGY-DENSITY CONVERTER FOR PORTABLE POWER (PDF)

POWER ELECTRONICS INTERFACE FOR A 100 W, 500000 RPM, GAS TURBINE PORTABLE POWER UNIT (PDF)

AN ULTRA-HIGH-SPEED, 500000 RPM 1 kW ELECTRIC DRIVE SYSTEM (PDF)

MICROFABRICATED SILICON CARBIDE MICROENGINE STRUCTURES (PDF)

HIGH POWER DENSITY SILICON COMBUSTION SYSTEMS FOR MICRO GAS TURBINE ENGINES (PDF)

Piston vs. Gas Turbine Engines, Efficiency, Final Thoughts

Pulling together all of the research and thinking I've done on the question of engine efficiency over the last couple of months, here is my conclusion...

If I were to write a design spec. for the power source for my "dream-team" agricultural robot harvester/tender it would specify a hybrid design consisting of a single-shaft gas-turbine generator combination driving a super-capacitor bank.

It would come in a drop-in package that could be changed in/out with a 1/2" socket set and 3/8" flare nut wrench.  The engine-generator set, the fuel tank and the super-capacitor bank would each be mounted as separate units.

But such a design is not realizable yet.  To date, there are no small 1-2 kW gas turbine engines commercially produced.  Though, the work being done on micro gas turbines makes one optimistic that someday they will be.

For now, if I had the machine shop to do it in, what I would do, would be to mount one of the Honda Gas Engines (showcased in a previous post) on a testbed and try to run it continuously for a few months and see how it holds up.  If it does, then it would make a good second option for an agricultural robot harvester/tender's power source.

If more information comes my way, I may yet change my mind on this subject.  But for now, I've gone as far on this subject as I can.  So it's time to put this aspect of robot design to bed for awhile and start on something new.

Piston vs. Gas Turbine Engines, Efficiency, Global View

The Global View  The issues of local and mid-range views of engine efficiency overlap the global view in the areas of maintenance and repair. 

Based on my experience, something that electronics and software engineers, working in the area of agricultural robotics, don't seem to appreciate is the high cost of machine maintenance.  Everyone is familiar with the high cost of auto repair these days, so why isn't the engineering community making the connection that robot repair is not going to be any less expensive?

Engine efficiency often comes at the expense of higher temperatures, higher pressures and greater mechanical complicity.  All of these are factors in decreased engine life, more stringent maintenance requirements and higher cost of repair.

Let's try some order-of-magnitude estimates, OME's, for the service costs of a robot harvester/tender and for the fuel budget savings for a 20% increase in engine thermal efficiency. 

First, an example repair situation...
   $100.00  2 hours labor
   $200.00  parts
   $100.00  tow-truck to pick-up/put-back our robot from/to the field
   $100.00  cost for a fill-in-for-the-day replacement robot
   $500.00  total cost to farmer for a typical repair  

Now look at the cost savings for a 20% increase in fuel efficiency.  Assume a fuel consumption of 2 gal/day.  At $4.00 a gallon, a 20% increase in fuel efficiency yields about 17% savings in fuel costs, which translates to about $1.35 a day.  At this rate it would take a year's worth of operation to balance out the cost of just one typical repair.   

From the point of view of the farmer's pocketbook, the bottom line is that reliability and low maintenance costs will always trump considerations of engine thermal efficiency .

Electromagnetic Actuators? Why Not?

Regardless of the novelty of a design concept, the laws of physics will always put limits on what any actuator can do. That good-old F=ma equation tells you that in order to move an actuator mechanism quickly, large forces are required to first accelerate it up to speed and then to de-accelerate it back to a stop.

If electro-magnetic forces are going to be the driver of an actuator's motion, then the equation F=iLBsin(θ) tells you how much wire(L), current(i) and static B-field you need to generate a given force level.

Given a possible 2000-8000 gauss range for permanent magnets, the currents necessary to generate any kind of usable force levels (with speed equivalent to human hand and arm motions) will be on the order of 10's of amps.  Then, with these current levels, the resultant resistive losses from the actuator's "windings" will generate copious amounts of heat that will have to be "dumped" somehow.

I worked on this concept a couple of years ago and for the above reasons, came to the conclusion that it will be impossible, using standard-technology electro-magnetic actuators, to recreate a robotic hand/arm system that mimics the action of a muscle fiber/tendon bundle.

Muscle fiber can only pull, not push. So to accomplish the finger, hand and wrist joint movements that we can, several muscle groups have to act simultaneously, pulling either together or in opposition to each other. Because of this fact, we have control over not just the position, force, and velocity of our hand movements, but also the overall mechanical stiffness of our hand/arm system as well.

Another thing to consider is the force level our muscles are capable of. If you look at our body's muscle-bone attachments and their attendant mechanical disadvantage in terms of forces and lever-arms, you'll see that our muscles don't just generate ounces of force, but 10's and 100's of pounds of force. And in the case of exceptional individuals, like an Olympic power-lifter for example, 1000's of pounds of force! I believe, the only thing in the same category, in terms of force per unit volume, is hydraulics.

The skeletal-muscular system that Mother Nature has evolved is truly extraordinary.

Now back to the topic of this post, if the goal is to create a robotic hand/arm system and do it using actuators that mimic the action of a muscle fiber/tendon bundle, then the following observations should be noted.

First, since a muscle fiber is a pull-only actuator, several different muscles, some in opposition to each other, have to fire smoothly, simultaneously and continuously together to reproduce the range of joint movements we have.

Second, the accuracy to which we can control our finger movements is tied to the rigidity that we can hold our hand/arm system to. There is no "brake" in the human hand/arm that allows it to be locked into a fixed position. The way we do this is to fire opposing muscle groups, in a full-on mode, and thereby force our hand/arm into a state of ridge tension.

These two observations together imply that for a robotic hand/arm system that mimics the action of a muscle fiber/tendon bundle, most, if not all, of the actuators will be in some state of "on"-ness all of the time. If one assumes voice-coil type actuators, operating in the 10's to 100's of watts power range, then multiples that power load by a factor of 10 to 30 (the approximate number of muscles in a human hand and forearm that might be engaged at any one moment), one finds that the total power load skyrockets. And of course along with the accompanying waste-heat cooling problem.

And this is just one hand/arm we're talking about.  Now multiply this by two hand/arms, then add in four to six legs for locomotion, then add in the power to run the harvester/tender's computer systems, and the power requirements for our robot become greater than any reasonably sized "portable" power source will be capable of delivering. 

Or to say it a different way; the reason factory floor mounted robots run just fine using electro-magnetic actuators is that they don't have to carry their power source with them.  Actuator power consumption is not a design constraint when your robot can plug into the power grid.  But once a robot is required to be self contained power-wise, then any excessively large current requirement will become a design show-stopper.  
 
Which is why, IMHO, that if a robotic hand/arm system is going to be made using actuators that mimic the action of a muscle fiber/tendon bundle, then it won't be done with electro-magnetic based devices.

Topic for a future post  The technological step I‘m looking for in hydraulics is to have a small fast-acting piezo-electrical driven pump that would mount directly onto the cylinder it is driving; thus forming a self-contained unit, a kind of "point-of-load" hydraulic supply.  This would get rid of the need for an external engine-driven hydraulic pump with its accumulator and rats nest of hoses that is typical of current hydraulic systems.

Piston vs. Gas Turbine Engines, Efficiency, Mid-Range View

The Mid-Range View  Fortunately for agricultural robotics most harvester/tender duties will be very even and predictable in their power requirements.  This makes a hybrid design a very workable choice for a harvester/tender’s power source.

Once the choice of a hybrid design has been made, the choice of actuator technology follows.  Since energy in a hybrid design is stored as electric charge, the most efficient pathway to convert that energy to mechanical work will be with electrical based actuators.  That is, once energy has been stored as electrical charge, it would be a detour in inefficiency to first use that energy to power an electric motor that in turn drives a pneumatic or hydraulic pump that then is used to drive the corresponding actuator type.

Topic for an upcoming post:  Conventional electromagnetic actuators have serious, even fatal, drawbacks when it comes to their use in agricultural robotics applications.  At this point, some thinking out of the box is in order.

Piston vs. Gas Turbine Engines, Efficiency, Narrow View

The Narrow View   With a few hours of web searching one can find a number of fascinating and far ranging discussions regarding thermal efficiency, covering everything from the theoretical realm of physics to the mundane world of practical mechanical issues.  

As an example of what is attainable in terms of thermal efficiency, here is a clip of a gas-turbine engine combined with a co-generation steam turbine that can attain a 60% thermal efficiency. 

Unfortunately for those of us designing robotics, large utility power plants, like the one featured above, enjoy an economy of scale that designers of agricultural robots won't have. 

One of the main design constraints for industrial/agricultural robotics is the KISS principle; that is, systems must be field serviceable by workers with a minimum of tools and expertise.  This means that designs for robotic power plants must consist of a minimum of moving parts.  Which means that the added complexities that come with co-generation schemes are off the table as design options.     

Fuel efficiency for an internal combustion engine is generally measured in  Brake-Specific-Fuel-Consumption.  This is a ratio of fuel burned to energy on the engine's output shaft and is given in units of pounds per horsepower-hour, or in units of grams per kilowatt-hour. 

The one aspect that is most important for our discussion is the concept of an efficiency “sweet spot” in the torque versus RPM graph; that is, a region in the torque versus RPM graph where an engine achieves its maximum efficiency [1]. 
 

BSFC for a Saturn DOHC engine with 20-HP operating curve marked in red.
Source Link

I've looked without success on the web for an equivalent graph for a gas turbine engine.  Based on the information I've come across, the "sweet spot" for a gas turbine would be smaller in area but higher in efficiency than a comparable piston engine; with efficiency dropping off dramatically as one moves away from that "sweet spot" region.    

Again, based on my readings, gas turbines have a higher efficiency over a narrow region in their torque/RPM graph, while piston engines have a "reasonable" efficiency over a broader area in the torque/RPM graph.  

In practical terms, if a robot has a narrow and well defined power requirement, then a gas turbine could be a good choice for its power source.  On the other hand, if a robot's power load is going to vary, then a piston engine will give the greatest efficiency when fuel consumption is averaged over all of the robot's working conditions.  This is one of the reasons that Boston Dynamics' "Big Dog" is piston engine driven.

The Hybrid Engine Concept  The idea is to decouple the engine from the machine it's driving by using an intermediate energy storage reservoir.  The engine drives a generator that charges either a battery or super-capacitor bank.  The robot then pulls its power from there.  This allows the engine to charge up its associated energy storage system while running at its point of maximum thermal efficiency.  It can then throttle back to a fuel saving "idle mode" when a sufficient buffer amount of energy has been stored-up.  This effectively decouples the narrow view from the mid-range view of thermodynamic efficiency.  

[1]. The "sweet spot" for the graph above is centered at a speed of 2500 RPM and a torque of 124.8 Nt-m.  The red 20-HP line represents the average power needed to move the Saturn car under typical driving conditions.  The closed curves represent lines of constant BSFC.  A lower BSFC means less fuel used for the same power output, that is, lower BSFC is equivalent to a higher fuel efficiency.

Piston vs. Gas Turbine Engines, Efficiency

What efficiency measures is a function of how broad or narrow one's view of a system's thermodynamic universe is.

A narrow view would be to look at just the thermal efficiency of an engine itself.  That is, what percentage of the chemical energy contained in an engine’s fuel gets converted to power output on the engine’s driveshaft.

A mid-range view would consider efficiency as the ratio of the work done by one’s robot to the amount of fuel its engine consumes.  In this case, one has to consider, not just the thermal efficiency of the engine itself, but also how efficiently that power is coupled to the work the robot is doing.  This is the impedance matching problem from before.

At a global view, the system's thermodynamic universe becomes an economic one.  That is, to the farmer efficiency is now the amount of profit generated per robot versus its cost of operation.  In the end, this is the most important figure-of-merit to the farmer.

It would certainly be advantageous if one could break down the question of engine efficiency into three such well-defined categories.  But such is not the case.

The first complication is that an engine's thermal efficiency is dependent on the power load and RPM it is running at.  This effectively couples the local and mid-range views together.  One can’t analyze an engine’s thermal efficiency without considering at the same time the mechanical system that engine is driving. 

The second complication is that the laws of physics put hard limits on the maximum thermal efficiency any heat engine can attain.  Ultimately what physics tells us is those higher efficiencies require higher operating pressures and temperatures.  These lead to higher mechanical stresses on an engine’s design, leading to more frequent breakdowns and required maintenance.  And it is in the area of maintenance costs where the local view of thermal efficiency crosses over into the global view of economic profitability.