Throttle control is for rotor rotation.

Yamaha's throttle control for rotor rotation
~ project developed in cooperation with Yamaha Motor Co in Japan and United Technologies Research Center in US
~ existing Yamaha 3.5m R-50 commercial rice-paddy-spraying helicopter used as the platform
~ three R-50s had their spray arms removed and re-engineered to accept new control and guidance equipment
~ R-30 is not a model helicopter; about 500 units have been sold, primarily for spraying small Japanese rice fields.
~ R-30 weighs about 100 pounds and has a 30 minute flying time.
~ this unit is normally flown by remote control via joysticks.
~ to master flying using joystick controller, takes several months and requires that operator be in constant visual contact with R-30.

~ a helicopter has 6-DOF in its motions, up/down, fore/aft, right/left, pitching, rolling, and yawing; there are 15 state variables.
~ se correspond to various flight modes such as forward flight, backward flight, sideward flight, hovering, hovering turn, vertical climb, etc.
~ a helicopter has a throttle control for rotor rotation and four other control inputs:
~ collective pitch for climb or descent by changing main rotor's lift by changing main rotor blade angle,
~ longitudinal cyclic pitch for forward or backward flight by tilting main rotor path plane fore or aft,
~ lateral cyclic pitch for left or right flight by tilting main rotor path plane,
~ antitorque pedals for heading direction control by changing lift of tail rotor through changing its angle

~ system is inherently poorly damped or unstable and is easily affected by temperature, humidity, wind, etc.
~ flight modes are cross-coupled; in addition to basic airframe rigid body dynamics, rotor and engine affect dynamics.
~ there are nonlinear variations with air speed, ground effects, etc
~ in addition, helicopters have many more flight modes than winged aircraft and are often forced to fly in difficult conditions (low altitude, near obstacles, etc).
~ vonventional control systems are based on feedback control, but helicopter pilots are able to operate in a predictive feedforward manner
~ to stabilize helicopter to compensate for time delays between change of input to output.

~ primary innovation in development of this system is two layer control system,
~ with voice commands entering at upper layer and control commands produced at lowest layer.
~ such a two layer system, which is highly modularized, allows command interpretation, navigation, and flight mode management
~ (take-off, hover, forward, turn) to be in top layer.
~ lower layer consists of four modular fuzzy controls for four control inputs
~ (longitudinal cyclic, lateral cyclic, collective pitch including throttle, and antitorque pedals).
~ each module consists of in-then fuzzy rules (39 rules are used for hovering, etc).
~ Sugeno has applied for a patent (1992) on aspects of this system, which he claims will also apply to other systems that move in 3 dimensions, such as underwater.

~ test helicopter is equipped with fuzzy controller, a command receiver, various sensors, a wireless camera and a telemeter.
~ TMOS 1000 sensor measures 3D accelerations, velocities, attitude angles, and angular velocities.
~ radio wave speed meter measures 3D velocities.
~ laser height meter measures height, and a magnetic azimuth sensor measures heading.
~ differential GPS system compensated by another GPS on fixed on ground measures 3D position with 30-100cm accuracy.
~ telemeter transmits all 15 measured variables to ground.
~ controller in an Omron FP3000 with an inference speed of 60 rules per microsec.

~ there are unmanned helicopters under development but these are controlled remotely by hand, like standard R-50.
~ Sugeno claims that automatic control has previously only involved hovering, maintenance of constant course and altitude after achieving stationary flight,
~ and automatic course control, with other control functions handled by a pilot, and that to date, no fully automatic control system has been developed.
~ using his system there is only one control and that is typically given by voice; operator need not be aware of helicopter's dynamics.
~ in fact, voice control is not a major issue in this project.
~ operator speaks a command into a microphone and digitized speech is then analyzed using a conventional speech processing box,
~ which decides if it is one of allowable commands and if so, displays its choice on a CRT.
~ if operator agrees with analysis he confirms choice by pressing a key, and command is then transmitted to helicopter.
~ a second microphone near speaker functions as an active sound subtraction system; operation environment is quite noisy.

~ Sugeno makes a persuasive case that fuzzy control can be effective in this complicated nonlinear environment,
~ but eventually it comes down to implementation.
~ In current demonstration Sugeno's staff showed three different flight sequences by speaking commands.
~ (1) Square flight with commands: take off, up, stop, forward, stop, right, stop, reverse, left hovering turn, forward, stop, land.
~ (2) Figure eight flight with commands: take off, up, stop, forward, turn right, turn left, stop, reverse, land.
~ Right circling climb with commands: take off, up, stop, left hovering turn, forward, right circling and up, stop,down, stop, land.
~ One other planned demonstration, image guided landing was not working properly and was not shown
~ (this involves searching via image processing of video image for a specified landing site).
~ Sugeno claims that this will work better when faster image processing hardware is mounted onboard.
~ (A new participant in project from UTRC, H.Winston, is a specialist in image processing and will work on this aspect.)
~ Everything that we were shown worked without incident. Sugeno has already distributed a simple video showing early experiments,
~ and he claims that a professional and updated version will be available by the end of March 1994.

Under development
~ Fuzzy commands such as fly forward a little, make a big right turn, etc.
~ Fuzzy fly by wire for the case when fine control with voice cannot be achieved.
~ Automatic landing based on image information (as mentioned above).

Future development
~ Automatic collision avoidance flight based on image and radar information.
~ Flight navigation system for out of view helicopter.
~ Automatic autorotation system in case of engine failure.
~ Sugeno felt that mission planning was the major research problem yet to be solved, i.e.,
~ from a high level specification of a task, develop a complete description of steps necessary to solve it --
~ "go find the ship that is on fire" (there is an analogous problem in robotics and other autonomous systems).
~ On the topic of commercialization, a Yamaha spokesperson felt that this system would be commercialized within five years.
~ current system hardware can be significantly reduced in weight with integration, increasing available payload capacity.
~ Sugeno commented that there are plans to initiate a company capitalized at about US$50M in near future,
~ which will involve several industrial concerns including UTRC and Yamaha Motors.
~ On a related note, Sugeno explained that Kawasaki Heavy Industries has just begun a 6 year US$100M program
~ to use this technology to develop a pilot-assisted system for a large helicopter.
~ If successful (and Sugeno believes that scaling up will not be difficult, but admits that he has no direct experimental evidence)
~ such a system could be an excellent transition to commercial applications without having to deal with more difficult problems associated with an unmanned unit.