mel siegel. Лекция по энергетике для робототехники

20140301 <[email protected]> energy issues in robotics 1

Will Power and Energy IssuesPut the Brakes on the

Engine of Robotics Growth?

Mel SiegelCarnegie Mellon University

School of Computer ScienceThe Robotics Institute

Sensing, Sensors, & Control Lab Director

MS in Robotics Technology Program Director


historically: sensors & human interfaces


recently: energy & power issues ...

liquid fuelbatteries

The Energizer (batteries) Bunny

“He just keeps going!”

The Energizer (batteries) Bunny

But in truth the Bunny's original animatronics engineer (Grant Imahara) reports that it actuallyran less than 10 minutes ...

“He just keeps going?”


batteries vs. ice creamASIMO runs for about 30 minutes on a

7.7 kg battery mass (out of 52 kg total mass),so requires ~400 kg/day of fully-charged batteries

an excellent 7.7 kg battery might store ~4 Mj

so Asimo's power demand is ~2 kW

compare with human power demand ~0.1 kW

would be provided by ~400 g/day of ice cream

~400 gm/day for human | ~400 kg/day for robot

so it is no surprise BigDog runs on liquid fuel!


what to measure / what to calculatescaling rules for geometrically similar vehicles: vehicle's characteristic linear dimension = h then on-board energy storage capacity ~h^3 baseline power requirement = P running time (T) on stored energy ~h3/P range (R) on stored energy ~h3v/P (v = speed) if P is mostly frictional loss to air or water then

P ~h2v, so T ~h/v, so R~h– which is why we use a few big

boats/airplanesvs. many small ones to travel long distances

(and similarly for any baseline power model)


partial conclusion - energy storage batteries are terrible compared to liquid fuels this problem is not unique to robotics:

– cell phones, electric cars, etc, are all battery-limited

potential ultimate solutions:– renewable electrodes?– fuel cells?– solutions that avoid storage:

• wireless transmission (while in motion)?• foraging for energy (in the environment)?

we're not green, but this isn't yet a barrier robotics can “ride the wave” of progress driven

by transportation & communication demands

also: energy from transient power sourcessunny?

cloudy?

solar panel

SCIENCE

ENGINEERING

INTELLIGENCE


what is the problem? some system ... mobile robot? bus? boat? ...

collects solar energy for immediate tasks and stores solar energy for future tasks

some requirements are discretionary ... do them now or do them later use an optimal mix of incoming & stored

energy the optimal strategy now depends on the energy

you predict you will collect later ! predictions can be based on a mixture of locally

collected data and downloaded weather data

“solar” and “weather” are convenient concrete examples, but same thinking applies globally


what are the questions? given weather forecasts today, can we predict

the energy we will be able to gather tomorrow? given our own local measurements of

insolation, other weather variables, sky temperature, etc, can we predict the energy we will be able to gather tomorrow?

what is the optimal approach to fusion of these multiple data sources?

plus some secondary technology issues, e.g., tracking the maximum power point of sources (like solar cells) with “interesting” V-I properties


partial conclusion - transient power it is mostly an information technology domain! hardware is generally well-developed

software to coordinate hardware components isn't

this is not a threat for robotics it is rather an opportunity for us to spin-off the

IT, AI, and communication technology advances in our smaller robotics context into the larger contexts of transient power availability and consequent energy storage requirements

even on regional, national, and international scales


overall conclusion ...Will Power and Energy Issues Put the Brakes on the Engine of Robotics Growth? at first it seems like yes:

– we talk about “bio-inspired” robots, but our energy storage media are far inferior to life forms'• the fix seems to be to gather energy locally• but management of transient sources may be

harder than path planning, etc, in robotics nevertheless, taking a closer look:

– energy storage is a crucial issue in fields much larger than robotics, so we can ride their wave

– transient power management is similar to optimization problems in robotics, so we can spin-off our progress into broader contexts


Another Topic - Briefly!


companies recently bought by Google(Wikipedia says “more than one/week since 2010”!)

DeepMind AI Nest home automation Boston Dynamics walking robots Bot & Dolly robot cameras Holumni robot wheels Meka Robots people-oriented robots Redwood Robotics robot arms Industrial Perception computer vision Schaft humanoid robots Flutter gesture recognition Makani Power airborne wind turbines


Albert Efimov asked this question:How will Google's purchase of ~10 robot/robotics companies affect the rates & directions of research, development, & adoption of robots/robotic systems?


Albert Efimov asked this question:How will Google's purchase of ~10 robot/robotics companies affect the rates & directions of research, development, & adoption of robots/robotic systems?

Mel Siegel gave this answer:The lion's share of robotics funding -- government and industry -- chases requirements articulated by the military sector; Google's fusion of many unique areas of expertise -- until now held by many small players -- and the funding that will now allow them to pursue their own dreams -- will trigger a huge explosion of products that will live or die according to their fitness to survive in the popular -- not just the military -- market.


Is this a topic for future panel discussion?


Будут ли проблемас питанием и энергией затормаживать темнироста роботтехники?

Will Power and Energy IssuesPut the Brakes on the

Engine of Robotics Growth?


Will there be a problem with the power and energy inhibit growth

robottehniki dark?

Будут ли проблемас питанием и энергией затормаживать темнироста роботтехники?