Importance of Autonomy?

XPRIZE

Should we require the teams’ solutions to be autonomous?

Proposed Prize (click "show" to view):

Wildfire XPRIZE - Rapid, precise and autonomous detection and suppression of wildfires to prevent the loss of life and assets

The Winning Team will autonomously detect and extinguish a spreading wildfire in a large, controlled area in 10 minutes or less.

The prize design:
Each finalist team will be assigned a 1,000 km2 (20X20 mile) grid. On test day, in that grid, XPRIZE will create several small, decoy stationary fires and one hazardous target fire. Once the fire is moving, or reaches 2 meters in diameter, the team will have 10 minutes to autonomously extinguish the target fire and any spot fires, while leaving the decoy fires untouched.

Why, or why not?

DanSelz

One reason that I think it makes sense to consider autonomy is that when it comes to the 3% of wildfires that grow into devastating fires, time is extremely precious and limited. Autonomous detection would seem to allow for more rapid, precise detection by eliminating bottlenecks with respect to human resources, capability, and availability.

Does this make sense to people? Other thoughts or POVs?

Utobou

@DanSelz I agree to you.
At the very least, I think it may be impossible to deal with large scale wildfire unless it is autonomy. Also, as you mentioned, the detection must be considered autonomy. After autonomy detection, the system should provide options for human judgment regarding countermeasures (fire extinguishing and fire control).
There are following cases to consider whether human should judge or let autonomy to take actions. The case with a wildfire in one place (probably in the case of the XPRIZE challenge), the case with many fires at the same time, the case there are no human beings or animals in the target zone, and the case where no destruction or damage is allowed in the target zone and so on. There may be various obstacles going forward for full-auto countermeasures in the real world like the target area so close to the borders (country, county, tribe, etc) is also another example to consider. Of course, in the worst situation, and you must protect the common greatest value in the woods without further delay, then you should push the full-autonomy switch immediately. I think this might be moral and ethical issues, which we are arguing on autonomous and general purpose AI.
In addition, I will be in trouble if I extinguish the wildfire without permission, such as slash-and-burn agriculture and natural wildfires that are necessary in the natural world. In such a case, even if there is an alarm, it is better that the final decision is made by humans. And I believe that it will eventually shift to autonomy in near future.
Anyway, as a natural event, wind speed, wind direction, combustibles (trees, etc.), temperature, humidity, soil water retention, terrain, day or night, weather ahead, value to be burnt, carbon dioxide emissions by the fire, resources needed for extinguish..., There are so many elements necessary for judgment for countermeasures, and the knowledge, experience and wisdom of skilled firefighters must be incorporated into the autonomy fire fighting methods.
Thank you.

akb

Innovation challenges often take great care to avoid specifying specific technological solutions - so that many (creative) ideas can be submitted, and the most effective solutions found during actual testing.

So based on that great philosophy, we would not say that solutions must be autonomous.

However, it can be envisaged that the most successful solutions might turn out to be those that are autonomous. The obvious reason being that fires can quickly turn into dangerous environments, in which the lives of firefighters would be in danger. This means firefighters would have to retreat from a fire, instead of completing the task to put the fire out. A robust (fireproof) autonomous technology could be left to stand firm in a dangerous situation and complete its mission.

Having said that, remotely controlled technologies can equally stand firm and complete their mission. These are not classed as autonomous: they might be manual or semi-autonomous.

Therefore, the challenge could accept all proposed solutions (autonomous or otherwise); provided realistic testing scenarios are put into place. This means testing the ability of a prototype, in all conditions, to: (a) access the land where the fire is [perhaps across difficult terrain and via man-made barriers]; and (b) to perform well at the location of the fire.

akb

I agree with @DanSelz point about detection being autonomous
- we do want reliable and rapid detection of fires.

akb

@Utobou makes a good point to. After autonomous detection of a fire, people may well be required to make a judgement as to what form of action (if any) is required, and authorised.

ymedan

I am also supportng a level of autonomy because not all fires are accessible to human operated equipment and sometime, the scope of fires is too large relative to the human resources available for fighting them. The "autonomy" should first be limited to executing a dynamic fire extinguishing plan, which has been pre-approved by a professional fire fighting experts.

andracretu

Regarding the Wildfires
Even the autonomous solutions could be, for example, drones with supertanks with water and other materials that could be useful for extinguishing the wildfires, the planes remain an immediate solution.
Very important in any situation, are technologies and AI for timing, synchronization and for the communication system, localization on the maps, radars, different confirmations that are needed.

Almog

I think the detection issue is successfully covered above. I'd like to focus on the challenge of stopping the spreading of wildfires’ extreme energy towards populated areas (not extinguishing). My argument is very simple, the only solution based on persistent kamikaze attack (10 hours and more, 100 tone of water per hour) aiming to dispersing water (absorbing the heat) by aerial platforms could work. Successful implementation of this approach brings more than 80% of the water accuratly, 10-20 meters above the fire front. These kamikaze attacks espacialy at night, can be done onle by unmanned aerial platform. The level of autonomy is depictrd in the [url=" https://youtu.be/d6DlRMM8Oa4"]video[/url].
Have I missed something?

Simonyu

Just received a patent allowance. Drone to fight fire without use water or chemical.
see www.RoboticVectorControl.com New concept to fight wildland fire.

SDhillon

Autonomy is a 2-edged sword, as a developer of an autonomy capable solution, the number one thing you here from fire services is "I don't want autonomous systems, no way I'm getting that." The number two thing you hear is "I don't have a budget to hire people to manage that." The point is that to satisfy the USA based fire services, you need a system that is somewhere in between Autonomy and managed in its main mode but capable of operating at either end of the spectrum.

akb

Good point @SDhillon
It's possible these days to develop robotics that's directed at the strategic level and autonomous at the tactical (or micro) level. I'd imagine we might want people involved in looking at relayed images of the scene a robot is about to tackle, and people then directing the robot to (semi-)autonomously extinguish the fire in a specific area (defined by a person).

Similarly, there is an option for avatar-robots where the person has full control of the robot/device. [e.g. in another context we have the proposed Astro-bot]