Law Enforcement Robotics and Drones – 5 Current Applications

Pamela Bump

Pamela is Managing Editor at Emerj. She previous worked in B2B digital publishing with Innovation Leaders and Boston MedTech. Pamela holds a Master's degree in Media Ventures from Boston University.

Law Enforcement Robotics and Drones - 5 Current Applications 1

In May 2018, Bard College’s Center for the Study of the Drone released a report estimating that 910 “state and local police, sheriff, fire, and emergency services agencies” had acquired drones in the United States by the end of 2017.

They also stated that this number had grown by 62 percent since 2016. Of the 910 organizations recorded with drones, Bard reports that 302 were county police or sheriff’s offices; 278 were municipal police offices; 293 were offices relating to fire or emergency management; and 37 were statewide agencies.

One government agency which has publicly acknowledged using drones is the Federal Bureau of Intelligence (FBI). In 2013, following leaks from former intelligence employee Edward Snowden, FBI Director Robert Mueller, said the organization had been using drones for domestic surveillance in a “very minimal” way, mainly in times of investigation.

Later, In 2018, following the Las Vegas shooting which killed 58 people, the LA Times reported that Indio, California police officers used drones to monitor security and traffic at Coachella, The drones reportedly hovered with a pre-determined route over the estimated 250,000-attendee festival and sent video to human police officers for review.

To explore the possibilities of AI-powered drones and robotics in law enforcement, we have put together a list of current government projects, commercial products and companies that claim to assist in surveillance and investigation processes with robot-related applications.

The following applications are split into two major categories:

  • Robotic Police Officers and Guards
  • Drone and Drone Software Applications

With this report, we aim to give law enforcement or security agencies insights on how others are using robotics and drones to prevent, investigate or fight crime. Below are five law enforcement robotics and drone applications, from surveillance to customer service.

Asia’s Approach to AI Guards

Dubai’s Police Robot

In May 2017, Dubai Police Department launched what it calls its first “robocop” at the Gulf Information and Security Expo. The rolling 5-foot-5 robot was built by Pal Robotics and commissioned by funding from the City of Dubai’s government.

The robot has been placed in high-population areas including malls since May 2017, shortly after the release, Dubai officials said the robot could primarily be used for surveillance and customer service purposes. When it was first announced in 2017, the robot was capable of the following tasks:

  • Greetings: The humanoid can say hello to people and answer directional questions, such as “Where is the police department?” in English and Arabic.
  • Touch Interface: With its touchscreen, the robot allows users to pay fines for traffic violations.
  • Facial Recognition: Dubai Police claim its facial recognition technology allows it to survey, record and greet faces that stand in front of its camera. The robot can also a remember an individual’s face and name in order to recognize them at a later time
  • Streaming: The bot reportedly records a live video feed which is sent to physical police office locations for human review.

According to the release, a civilian or tourist can tap a touch screen on the bot’s chest and follow prompts to digitally pay a traffic fine, report a crime or enter a chatbot interface where they can ask questions. By the time it was placed in malls, UAE officials, said its vocabulary had been expanded to add phrases in Spanish, Russian, French and Chinese languages.

In the release, Dubai officials say they hope this robot will be one of many which could make up one-quarter of the force by 2030. They claim that they aim to build an amplified version of the technology and hope to create the world’s biggest policing robot within the next two years.

This large robot would have similar features to the basic humanoid, but would also have stronger wheels with the ability to pursue suspects at a maximum of 80 kph per hour, according to the release. They also say they hope all of the developed robots, including the “robocop” will be able to record parking violations.

According to the press release, these robots were created to lessen the workload of customer service and other small tasks of human officers. Dubai Police’s Smart Services Director General said in the release that residents and tourists regularly visit police offices for questions and other customer service needs. This press release does not specify whether they are aiming for this robot to be a humanoid, or if it could look more like a robotic police vehicle.

“Most people visit police stations or customer service, but with this tool we can reach the public 24/7. … It can protect people from crime because it can broadcast what is happening right away to our command and control centre.” – Brig Khalid Al Razooqi, Director General of Smart Services at Dubai Police (May 2017)

In the 3-minute video below explains the purpose and abilities of the robot:

While Barcelona, Spain-based Pal Robotics does not seem to have created a similar policing robot for other countries, the company claims to offer similar “humanoid” robot assistants for various customer service and other industries. Dubai’s robot officer is a customized version of Pal’s REEM model.

The 1-minute video below demonstrates how Pal claims the robot can be used for navigation and customer service assistance:

A list of clients and other use cases for this company could not be found on its site. The company, founded in 2008, has just over 100 employees according to its Linkedin page. It is not clear from its website, Crunchbase, or other sources of how much funding it has received.

The cost of purchasing and customizing the robot was not noted in the press release or on the Pal Robotics website. Pal also does not disclose pricing information for any of its models, but does note that they are available for purchase or rent.

South Korean Prison Surveillance

In an effort to lighten the workload of human guards, South Korea has pilotted two guard systems including its previous prison guard robot and an ongoing drone pilot, according to the Korean Times.

In 2011, South Korea’s Asian Forum of Corrections announced that it had commissioned and began piloting robotic guards to survey prisons.

South Korean officials told outlets including the Korean Times that the 5-foot robots would:

  • Autonomously roll on built-in wheels around a pre-set path on prison grounds
  • Use motion and thermal sensors, as well as a built-in camera, to detect and begin recording what it identified as “unusual behavior”
  • Notify and feed video to a prison control room when a possible incident is detected

While a press release from the South Korean government could not be found, Reuters noted that “unusual behavior” could include noise, high thermal energy and violent movement patterns, such as running or quick body movements.

Reuters also noted that South Korean officials believed that this type of robot may help with staffing assistance and limiting human workloads in low-resource prisons.

“The purpose of developing this kind of robot is to secure the prisoner’s life and safety and decrease the workload of corrections officers in a poor working environment.” – Lee Pack Chol, Asian Forum of Corrections, (Reuters, April 2012)

In early reports, The Telegraph estimated that the project would cost just under 800 million won, or around $717,000.

The video below from CBS demonstrates and explains how South Korea claimed the robot could identify and notify guards of unusual behavior.

The robot was piloted through 2012, but was later scrapped by the government due to privacy concerns and unspecified “technical” issues, according to the Korean Times. In July 2017, the publication reported that South Korea had begun a six-month pilot where drones would be deployed to survey the inside and outside of three prison facilities.

For the pilot, the Times said that South Korea’s Ministry of Justice purchased four drones for a total of 60 million won, or $53,600.

Further news has not been reported on whether or not the pilot was a success or is still ongoing.

SGR A1 for Korean Border Patrol

In 2010, South Korea deployed border-guarding SGR A1 robots. Reportedly built in partnership with Samsung Techwin and Korea University, this robotic surveillance camera commissioned by South Korea sits on the South and North Korean border.

The IEEE reports that the robot is capable of:

    • Using computer vision and thermal cameras to detect moving subjects
    • Turning its camera to follow moving subjects
    • Firing autonomously at a subject when triggered with a built in 5.5mm machine gun and a 40mm grenade launcher

The number of units stationed on the border has not been reported.

According to its demo video, when the SGR A1’s sensors spot someone walking across or toward the borderline, the static robot and its cameras turn to follow the subject. It then sends an alert and video feed to a nearby South Korean control center. While the machine can be set to fire on its own at any subject when triggered by them crossing the border, a soldier watching a video feed in a control room can also choose whether or not to fire.

While there were no videos available directly from the South Korean government or Samsung Techwin, the two-minute video below demonstrates how the camera could reportedly work.

Knightscope

Airports, malls, stadiums, casinos and other private but densely populated businesses have used Knightscope, a commercial security robot. The company, which calls its Knightscope models “autonomous data machines,” claims the dome-shaped robot can:

  • Respond to civilians’ verbal or text comments with pre-recorded messages
  • Uses machine vision to detect people and license plates
  • Captures thermal imaging
  • Geotag and mapping its location using GPS and laser technologies
  • Use four built-in cameras to record and survey the surrounding 360 degrees
  • Navigate autonomously to a charging pad when it is low on battery
  • Allow users to contact authorities through its touchpad

A 2017 report published by UC Davis Law School noted that Knightscope was also in the process of developing weapon recognition technology for the robots which can identify shapes like that of a gun or knife. They did not note what other weapons might be recognized by the robot.

In the 5-minute video below, Knightscope VP of Marketing and Sales Stacy Stevens notes the key differences and shared AI-based similarities of Knightscope robot models:

Knightscope does not list clients on its site, but organizations including Carnegie Mellon University, LaGuardia Airport, as well as Twin Arrows Casino and Resort say they are currently using the robot.

Knightscope’s Crunchbase notes that the company, founded in 2013, has received $46.6 million in funding. In 2018, the company released an open letter saying it would offer $500,000 in services to a college campus that submitted the most “compelling letter” about how autonomous robots could improve security in a school setting.

Knightscope’s Senior Vice President and Chief Intelligence Officer, Mercedes Soria, received an advanced degree in artificial intelligence from the Massachusetts Institute of Technology.

Drone and Drone Software Applications

DJI

China-based DJI, founded in 2006, has sold its products to clients including the European Emergency Number Association (EENA), as well as US Fire departments in Florida and Georgia. Local news outlets have also reported the use of DJI’s Phantom 4 within police departments in Connecticut and Washington D.C.

DJI has released various drones, including autonomously flying Mavic, Inspire and Phantom 4 drones. The commercial drone seller has stated that its new Phantom 4 model uses “smart technology” like vision processing to detect objects in its path and avoid them. The company claims it can also hover when a human releases controls. If a flight is “disrupted” by an object or human’s inactivity outside of autopilot, DJI says the drone will hover back to the location where it launched.

While DJI says its machine vision can be used to help the drone identify and avoid collision objects in the air, the company claims its software is also accurate enough for the drone to hover safely indoors.

DJI says the Phantom 4 also has an “Activetrack” feature which allows a user to watch the flight from the camera’s perspective on the DJI smartphone app; put a square around an object, such as a person or animal; and tap a track button on their personal device which commands the drone to follow the subject.

This 3-minute DJI tutorial demonstrates how the company claims its Activetrack feature works:

The US military claims to be a former client of DJI, previously using its autonomous drones on the field. It is not clear what missions the drones were used for, or which models were purchased. However, In 2017,

A US memorandum banning the military use of DJI products was leaked by the drone news site SuasNews.com. The document stated that the China-based product, which sends data to its app, could be capturing surveillance data from the US or military which could then be used against the country.

One study from Kivu reported that DJI was not collecting and feeding data, however, this company was commissioned for the study by DJI. From our research, we could not find any further evidence showing that the suspicion was proven through a trial or investigation.

Following this ban, DJI announced a new privacy mode which they say can prevent certain data from being transferred online.

According to DJI and Bard College, the drone company has also commissioned studies on military and civilian drone usage with the college’s Center for the Study of the Drone.

In April 2018, DJI released a report, with data gained from outside news outlets, claiming that drones including their own models, had assisted police in saving 65 lives within the previous 12 months. While the number of autonomous or AI-powered drones involved in these rescues is unclear from the report, it notes that drones with thermal imaging cameras, “saved the lives of at least 15 victims who were hidden from view by darkness or obstacles.”

The final pages of the report include a full list of hyperlinked news stories used to add up the report’s results. It was also not clear which life-saving events were directly linked to DJI.

EENA says it began a collaboration with DJI in 2016 to see how drones could be implemented into public safety protocols. While EENA’s press release did not give specific numbers on the lives saved in the last two years, it linked to a video that stated one life was saved on a search and rescue mission with help from a DJI drone.

The video noted that first responders could not find a lost hiker in dangerous conditions on a mountain in Iceland. The drone recorded video of the area as the traveler as he was signaling with his phone light. This and drone location software allowed the responders to find and report to the person’s location.

This video case study did not note whether any AI drone features were used during the mission, but the press release, published in April 2018, said that due to accomplishments like this, EENA would be continuing research with DJI.

DJI also recently partnered with Axon, a law enforcement surveillance product company, on a project called Axon Air. Through Axon Air, Axon, which sells products to law enforcement offices, will also begin selling DJI’s Phantom 4, Inspire 1 and Matrice 210 drones, according to the release. The announcement also notes:

“The Axon Air program allows law enforcement agencies to purchase drones from a trusted partner and links DJI’s drone technology with Axon’s connected data network and Evidence.com services – the same platform that more than 200,000 public safety professionals use today.”

DJI’s leadership includes CEO and Founder Tao Wang, who studied at the Hong Kong University of Science and Technology. According to LinkedIn, its team also includes a number of researchers and engineers who have studied computer science and robotics, including Rohit Sant, a Software Engineering Manager with a Master of Science in Robotic Systems Development from Carnegie Mellon University.

DroneDeploy

California-based DroneDeploy claims its software links with drones from companies like DJI. DroneDeploy claims its technology can assist law enforcement at accident scenes with its 3D mapping feature.

When integrated into a compatible drone, the company says its software can:

  • Program a drone to automatically hover on a pre-set route
  • Photograph an aerial view of a scene
  • Take measurements of a scene based on geolocation sensors and the images
  • Identify colors and objects, such as cars, in its images
  • Auto-generate a 3D map using the measurements, colors, and objects it identifies.
  • Generate a live thermal map in flight

According to DroneDeploy, users wishing to use their drone for 3D-mapping can download the company’s app on an Android or Apple device and follow instructions to connect the software to a drone. For DJI drone users, DroneDeploy advises that they close the DJI app during the DroneDeploy app setup. Once connected, a user can enter the DroneDeploy app, see a satellite map of where they are and map a drone route with their finger.

When a user taps a begin flight button, the drone will fly on the requested route and take photos. DroneDeploys adds that photos can then be uploaded on DroneDeploys site which can stitch together and return a 3D map to the user, according to the company.

In the video below, a DroneDeploy user demonstrates how to 3D-model a property using the software partnered with a DJI drone. Along with 3D-mapping, the video also notes how DroneDeploy allows a user to designate a path so the drone can fly autonomously.

While the video states that it may take a “few hours” to create maps based on the drone images, DroneDeploy claims to offer a Live Map feature which can show a generated map on a user’s mobile device.

DroneDeploy also announced in June 2018 that it had launched a real-time Thermal Live Map feature, which they claim could help first responders find lost individuals. Similarly to its Live Map feature, When using this feature, DroneDeploy claims the heat-oriented map will show up on user’s smart device while the drone is in motion. Similarly to a weather map, dark red areas show the highest temperatures, while yellows to blues show average to extremely cold temperatures.

The 1-minute video below shows how a construction worker can use DroneDeploy’s Thermal Live Map feature to see areas of hot and colder temperature around them:

According to DroneDeploy, the software, coupled with a Mavic Pro DJI drone, was used by the Harrisville Police Department in Alabama. Following a traffic accident, a remote officer in a police department used the drone to photograph the surrounding three acres on a 7-minute flight. With this image and location data,

DroneDeploy states that its software was able to geotag each image and rebuild visuals of the incident in 3D. They say this gave investigators more insight on what caused the accident. This also allowed the police station to cut down on investigation time, according to DroneDeploy.

DroneDeploy’s technology was also used for disaster relief purposes following Hurricane Harvey. DroneDeploy wrote that its software was used to help the non-profit group, Humanitarian Drones, to map the areas in Houston, Texas to further examine the damage.

While other law enforcement clients could not be found on the website, DroneDeploy has a variety of customers in construction, disaster relief, first response, and agriculture.

The company CTO and Co-Founder, Nicholas Pilkington, has a PhD in Computer Science from the University of Cambridge and has previously held software engineering roles at Toshiba and NVIDIA, according to his LinkedIn profile.

Takeaways for Law Enforcement Officials

While robotic applications for law enforcement in Asia were created or commissioned by the government, DJI’s autonomous drones seem to be the most popular tool in the US, with a history of use in the military, police departments and first response.

The ground-based robotic applications above claim to have the similar goal of conserving police or security guard staff, as well as time-related resources. While Dubai claims its robot will focus on customer service and low-energy surveillance by answering civilian questions and recording video streams, companies like Knightscope similarly offer customer-service-oriented security robots which can also allow users to ask for directions, date and time.

South Korea reports its prison warden was implemented to avoid the need for time-consuming basic surveillance tasks, such as walking around prison grounds to ensure that prisoners are behaving. Similarly Knightscope’s clients — which have based the robots in high-populated areas like airports, casinos, and colleges — similarly note that they are looking for added security, safety and surveillance options for their facilities.

Many applications have also been adjusted to address growing privacy concerns. When drones began getting noted for their use in federal investigations, the FBI was questioned by Congress about the application’s level of surveillance on domestic citizens. Additionally, while DJI announced the development of a “privacy mode” following its US military ban, South Korea sited prisoner privacy as a reason for ending a pilot with its prison guard robot.

Drones vs. Robots in Cost Efficiency

It seems that some governments may be turning to drones for situations like surveillance, while they may turn to humanoids to lessen the smaller tasks of human police force workloads.

While police in Dubai, located in one of the world’s wealthiest countries, plan to expand their robotic force, South Korea did not try to create a new surveillance robot following the failure of its prison guard pilot. It instead purchased a fleet of drones that cost 13 times less, at 60 million won, than its original robot, which reportedly cost over 800 million won to build.

While we could not pin exact cost numbers for Dubai’s robot or Knightscope models, both robot companies offer units for rent as well as purchase.

US Use of DJI and Potential Competition

While we see law enforcement robotics initiatives in other countries, developments in the US seen to be more focused on drones for the time being.

Despite fears of security and privacy breaches, officers and the FBI have announced the use of autonomous drones which can fly and take photos on a set path. Reports from DJI, as well as local and large news organizations, say that the company’s AI-powered drone models, such as its Mavic and Phantom, have been popular among police departments and the U.S. military.

Although DJI is reportedly a top-selling drone company, other startups have released drones with similar features. As the government considers new options for drone usage, another leading drone providers, such as Parrot, or GoPro, could gain more business from law enforcement agencies, Additionally, some other drone startups, like Skydio, claim to similarly offer machine-vision features.

 

Header image credit: Pixabay

Subscribe
subscribe-image
Stay Ahead of the Machine Learning Curve

Join over 20,000 AI-focused business leaders and receive our latest AI research and trends delivered weekly.

Thanks for subscribing to the Emerj "AI Advantage" newsletter, check your email inbox for confirmation.