AUTONOMOUS ROBOTS FOR LAW ENFORCEMENT

One of the greatest concerns facing officers across the country today is the threat of a meth lab. Meth labs produce byproducts and gases that endanger life and property, and law enforcement officers as first responders must enter these threatening environments in the course of doing their jobs.

When a meth lab is suspected, officers first must determine if a lab truly exists. If so, when officers execute a warrant they may need to wear one-use sealed contamination suits if the lab is thought to be in a dangerous cooking mode where cyanide gas might be produced. These suits are relatively expensive but also cannot be guaranteed to be tear-free once having been put on. This, in turn, creates a potentially fatal safety hazard for an officer.

But imagine if "thinking" robots made the first entry into these hazardous surroundings, while responding officers watched their activities from a safe distance away.

By using autonomous robots, law enforcement could gather intelligence in a stakeout and have the robots enter the area before sending in a SWAT team.

In a stakeout, robots could covertly observe a building or situation to watch for anomalies, suspicious activities or even sense heat patterns. An unattended robot could inform a base operation when an exceptional condition occurred. Or, when used in conjunction with an officer, the robot would be able to continue an observation when the officer was off-post or occupied with something else.

The robot, or a team of robots, also could be commanded to investigate the suspected meth lab. The robots, operating independently but as a networked team, would then approach the lab with each robot sharing its "viewpoint" with the other robots as well as any observing officers.

By sharing information, the robots multiply the effect of being a team, enabling a more effective and exhaustive investigation in a shorter period of time. Because these robots think for themselves, human manpower can be devoted to maintaining overall situational awareness, which improves safety for the entire team.

As robots enter a lab's vicinity, they use sensor arrays to determine if any humans are present, and whether the individual's are armed or extremely agitated, and potentially dangerous. The robots also can determine whether chemical or dry cooking is taking place, and if poisonous gases are present in the air. The robots use efficient algorithms to search each room — having a plan for every type of contingency. In some situations, they would be programmed to alert and acquire the assistance of officers.

This scenario illustrates the improvements in safety and productivity the law enforcement team may experience by deploying "thinking" robots. In effect, the robots are really a force addition, enhancing on-ground capabilities without requiring human operators.

Autonomous robotics
What is autonomous robotics anyway?

The most succinct way to describe it is to compare autonomous robots to remotely operated ones, such as an EOD (Explosive Ordnance Disposal) robot. With an EOD device, the officer controls the robot's movement through macro-type control bundling that allows sequences of control options to be invoked. With autonomous robotics, however, the robots are commanded to do tasks but in general, are not controlled while performing the task. An officer might be able to override an autonomous robot's actions, but in most cases, he would simply give the robot directions for it to execute autonomously.

Over the last 20 years, the area of robotics has evolved through three generations of capabilities as shown in Table 1 on Page 70. Each generation of "intellect" was focused on some type of task or groups of tasks with a typical example shown. EOD, for instance, is basically a hazmat-type task and only requires sensor feedback, which is the hallmark of remote-operated designs. The current generation of law enforcement robots, including EOD robots, is second-generation units.

Third-generation, or autonomous robots, recently became available. These systems are capable of making real-life decisions and acting upon them. Near-term uses for these devices would be in the military but since police work is very similar in nature, the products will soon find their way to the law enforcement market. Currently, there is significant military-funded research into autonomous robotics that is related to police work. Thus, it is time for law enforcement to seriously discuss the initial range of applications for autonomous police robots.

The "thinking" robot's role
Autonomous police robots have always been thought of as, first and foremost, improving officer/community safety while enhancing operational efficiency. However, any positive or negative impact this technology would have on law enforcement is more driven by the roles the robot would fulfill within police work. After having already made use of EOD robots and other technology, it becomes clear this technology fits well within two restricted roles in regards to autonomy of operation: customer/citizen information service and officer assistant.

An autonomous robot acting as a mobile citizen information center would link the customer to a second-level human help desk. This beneficial public service application would also help save patrol manpower.

If an autonomous robot is used for police tasks then it must act as an assistant to the officer. Consider how K-9s, for instance, are used within police work today. Autonomous robots will in many ways be treated similar to police K-9s, thus leading to the designation as an assistant. In a broader sense though, the robot is a more versatile assistant able to perform "smart on-demand" laboratory analysis, intelligently interact with information sources on behalf of the officer, record a permanent yet easily retrieved history of events, etc.

This latter role as assistant does have a substantial impact on the autonomous robot's daily operational mobility requirements. The robotic assistant would need to be available to the officer nearly all the time and be mobile outside of a vehicle in order to be truly beneficial. In dismounted mode, the robot will need to be capable of traversing typical urban or rural environments. Initially the frequency and range of environment may be mitigated since mobile locomotion and power considerations constrain the types of missions where the robot will be effective. In general, the duration of a dismounted activity will be from 30 minutes to 4 hours, depending on the mission, before a partial recharge would be necessary.

The role as assistant also carries significant legal impacts and limitations that must be understood and accounted for. An autonomous robot could create a substantial amount of inadvertent damage if allowed to roam freely in all situations as all contingencies cannot be accounted for. In addition, because the robot is akin to a portable laboratory it will held to the precedents already in place for breathalyzers and other similar devices. For this reason, using robots at a traffic stop for drug or alcohol detection does not appear to be that useful due to legal restraints. Here, a robot cannot be used unless probable cause is evident. The same would be true of using the robot to "see" through the car to detect weapons, as this too would be a violation of privacy laws.

Beefing up officer safety
Officer safety itself is an extremely important area for an autonomous robot application. A simple application is one of "calling home" when an officer is injured and providing up-to-the-moment status on the officer and situation to linked teams. Some discussions have centered on using the robot to protect the fallen officer but concerns arise about how the device would handle a civilian trying to help the officer. How will the robot know the difference between someone wanting to help and someone trying to hurt?

Arming autonomous robots with lethal or less-lethal weapons, allowing them to use force or be allowed to move at-will could help protect the officer. However, it also could cause injury to someone or damage to property. These types of activities would be best left for the future after a good amount practical field use has occurred.

Building intelligence
One of the most important applications for autonomous robots involves intelligence and information gathering, namely searching, scanning, detection, recording and surveillance. These capabilities are largely achievable in the near future.

The activities of searching, scanning and detection offer their greatest reward in improving officer safety. But legal restraints, however, will always inhibit the usefulness of robots in searching or scanning for something, particularly if there is no warrant. Even if there is a warrant, other than when investigating a hazardous environment, such as a meth lab, most robot searches would not be very useful or an improvement over human searches. Even so, the ability to "smartly" scan in order to keep the officer safe from harm seems to be a good area for concentration.

"Smart" scanning refers to the ability of the robot to know when and where to scan, to avoid providing information protected by privacy laws. The officer would require the ability to command a scan or override one.

"Smart" scanning would be most useful in situations such as a domestic disturbance where the officer must separate the parties but also needs to put them in "safe" areas. Here the robot would excel at identifying potential problems or providing assurance of a protected area sweep. Alternatively, using the robot to rapidly search large complex areas such as a warehouse also shows its positive attributes.

The thinking robot's ability to access public area video cameras and databases, providing instant online capabilities and giving a broader view in terms of both information and sight, is beyond comparison with what is done today. The robot's autonomy allows it to intelligently peruse these data sources without requiring an officer's attention unless something vital is discovered.

Employing such robots for surveillance and recording tend to improve productivity and accuracy more so than safety. Being able to deploy a robot to stakeout a location and intelligently notify an officer of a notable situation is a tremendous boon to productivity. Furthermore, the ease of concealment and long periods without attention make the use of robots for surveillance a superior solution.

Recording would apply to both surveillance as well as general activities. While assigned to the officer, just like is done for vehicles during a traffic stop, such robots would be able to record activity upon command or once an event trigger occurs.

In squad cars, when the officer activates the lightbar the recording system turns on. For a robot the triggers would be quite varied but basically would operate in a very similar fashion. Having a video and audio time-stamped record would aid in creating reports, providing evidence, and also cutting back on nuisance items such as citizen complaints.

The potential for direct linkage into public area video systems would apply to surveillance as well as searching and scanning as previously noted. Here the robot is able to get a bigger picture and thus able to be more productive in its observations.

With today's high-tech criminals, officers need more tools to keep them operating safely and effectively. But they do not necessarily require more tools needing individual operation. In fact, in the future, the largest improvements to officer safety and productivity will likely be driven by tools, such as autonomous robots, that seem to "think" for themselves.

Joseph Weiss has held various engineering, general management and business development positions at RCA, GE and ARINC. He has extensive experience with architecting, marketing, winning, developing and producing information technology intensive systems for mission-critical use. Currently he is a technology and market research consultant for the federal, state and local marketplaces and is associated with the market research firm EMSI out of Alexandria, Virginia. He can be contacted via e-mail at JFWeiss@emsiusa.com.

TABLE 1 - Robotic generations
1st - 1985 2nd - 1995 3rd - 2005
Intellect Repetitive Programming Sensor Feedback Decision Capable
Typical Task Assembly Hazmat Military

Loading