Are perfect robots a dream? Robotics has found its niche in many companies and is continuing to evolve. But many times, they end up failing! This could be because of technical glitches. On the other hand, it could be because their purposes were not clearly stated by the companies. The robotics industry is evolving at an alarming rate. Robotic industry successes are far too ahead of time. But arguably brilliant robotic industry researches offer valuable lessons about robotic technology. This helps to eliminate glitches and lends a helping hand for companies to succeed and go mainstream.
The robotic industry failure rates are more than 90% as per studies and discussions! This is a clear offshoot showing that robotic industries require expertise and synergy going forward!
Most-hyped Robotic Failures
Everyone recollects how Honda’s ASIMO robot faced difficulty on its very first day at work. The robot had problems understanding human gestures. It failed to recognise the difference between visitors raising their hands to ask questions of people raising hands to take pictures. This occurred on the very first day of its work at a museum guide in Tokyo.
Another tragic incident was that of the robot ‘Mitra’. Designed and developed by a Bengaluru-based company, the robot was programmed to welcome the Indian Prime Minister, Narendra Modi and Ivanka Trump at the GES (Global Entrepreneurship Summit) at Hyderabad. Mitra failed to perform, the reason is poor coding. It simply could not handle multiple requests simultaneously. As Modi was requested to press the Indian flag button, Ivanka Trump also ended up pressing on the US flag button more or less at the same time. Mitra could not function properly because of the confusion due to overlapping requests.
Most robots that are ‘technology dinosaurs’ at present have stories of failures in their early stages of development. These failures foretell that there are a lot more to learn about robotic technology from pitfalls and experience to succeed! Despite strong researches, robotic industries are still known to be error-prone with regular failures.
Are glitches the reason for Robotic failures?
Information on how and when robots fail helps in identifying the drawbacks in the current robotic industry. This helps the robotic industries to fight the challenges they face in developing fault-tolerant control robots.
Recent studies of robotic performance in USAR (Urban Search And Rescue) and MOUT (Military Operations in Urban Terrain) have shown a significant lack of reliability!
Factors affecting robotic performance are plenty. Categories of failures depending on the source of robotic failure are classified into two, namely :
- Physical Failures: These failures include those that arise due to common physical systems found in all robot platforms. These are effector, sensor, control systems, power and communications. Examples are wheels, motors, grippers and treads.
- Human Errors or Failures: These are also called errors. They occur due to human interaction with the robot or the computer.
Basic statistical analysis done to identify reasons for physical failures in various robotic industries are based on several key factors:
- MTBF(Mean Time Between Failures)
- Average downtime
- Frequency and Impact of Failures
- Repairability
- Dependability Computing
- Effectors and Sensors
- Control Systems
- Power and Communications
- The manufacturer provided software or any other remote OCU(Operator Control Units)
- Availability Measures
Human Errors or failures occur due to interaction issues and are further classified as
- HCI (Human-Computer Interaction) or Mistakes: These arise due to misunderstanding the situation.
- HRI (Human-Robot Interaction) or Slips: These arise where the operator attempted to do the right thing but was unsuccessful in performing.
The severity of each failure is evaluated by two attributes, namely :
- Repairability
- Impact
Failures are categorized based on the impact of a failure that occurs while they perform their mission. A terminal robot failure terminates the robot’s current mission. Whereas a non-terminal robot failure introduces some noticeable degradation of the robot’s ability to perform its mission. Repairability is yet another factor that has a major impact. A field-repairable robot can be repaired under favourable environmental conditions with certain common equipment. Whereas, a non-field repairable robot cannot be repaired in that manner.
Researches state that physical failures occur on an average, once every 24 hours and human failures occur every 17 minutes of robot usage time. Statistical analysis showed that the control systems were the most common source of failures and showed 32%!
How do you recover from Robotic Failures
Robots have evolved over the years into many streams. Emerging trends in robotics have led to crucial problems. For instance, programming robots were easy when the tasks involved were just picking and placing simple objects. But recent developments and requirements are not based on simple needs and tasks.
Robotic programs require strategies to detect, analyse and prevent potential catastrophes. Since many robotic actions are irreversible, researchers need to focus on technical know-how and their experience for success!
Automatic Error Recovery has not been fully addressed as yet! This has been difficult because of the lack of know-how about the physical environment in which the robot operates. Hence, automatic error recovery plays an important role in robotic industries.
Robotic Risk Assessment and Reduction
If a robot could perform surgeries and cure diseases, the robot would have been a technology jewel. Wouldn’t it? But the reports state that among 1.7 million surgeries carried out between 2007 and 2013, there were 144 deaths, 1391 injuries and 8061 device malfunctions during robot-assisted surgeries. This was reported by the US FDA(Food and Drug Administration). Another incident is when a robot failed to stop going forward even after it failed to diffuse a bomb. This is where a collaborative robot could find a fit. The best solution could be if a robot can sense human actions and start/stop respectively!
According to ISO 12100, the risk associated depends on two factors namely :
- The severity of the harm
- The probability of occurrence of that harm.
Risk assessment is required for the design, integration, installation, testing, verification, operation, training and maintenance. Designers are required to carry out risk assessments.
Equally important is the need for risk reduction. This is necessary to implement protective measures. The ISO 12100 states a 3 step approach for risk reduction, namely :
1. Inherent safe design measures (for hazard elimination).
2. Safeguarding and complementary protective measures (such as fixed guards, movable guards with interlocks and safety devices).
3. Information for use (such as safe working practices for the use of machinery, warning of residual risks and recommended personal protective equipment).
To ensure robot safety, specific risk reduction strategies are necessary for verification of the safety requirements to be taken by robot manufacturers, listed as follows:
Analysis :
- Visual inspection
- Practical tests
- Measurement
- Observation during operation
Safety requirements are necessary to be implemented to ensure robot safety and thus reduce risk of robotic failures. Implementing well-tried safety components including fault monitoring and basic safety principles are essential.
Requirements :
- General requirements(for instance, fixed and movable guards)
- Activating controls(Example, status, indicator light, pendant)
- Safety control(Hardware and software)
- Robot stopping functions(Example, protective stop functions, emergency stop functions).
- Reduced speed control
- Operational Modes
- Pendant controls
- Control of simultaneous motion
- Collaborative operation requirements
- Singularity protection
- Axis limiting
- Movement without drive power
- Provision for lifting and electrical connectors
Implementation:
- Implementing well-tried safety components including fault monitoring and basic safety principles are essential. Examples include the following :
- Using safety position switches with forced opening of their contacts and mounted positively to monitor the position of movable guards.
- Preventing modification to the program when electronic programmable systems are used to control safety functions.
- Use mechanically linked safety relays (not ordinary relays)
- Use safety light curtains (not optical sensors)
- Separating safety control and operation control and hence decreasing the chances that unwanted modifications cause by mechanics, electricians, and programmers.
- Protecting safety position switches and safety devices from harsh environments which could degrade them and result in premature wear and damage.
Saving the best for the last, the speed of the robot end effector must be controllable at selectable speeds and under reduced speed control. Enabling these results in a protective stop of the robot!
Review :
Review of application-specific schematics, circuit diagrams and design material
Review of task-based risk assessment and
Review of specifications and information for use.
Get Automated
Applications of robotic automation are already in place in every industry. The time is “Now”! Get started right away with our customized product and service-based solutions.
Embrace the future. Contact us to save your time and money to deliver a quality product.
