Sunday, July 23, 2017


Robotic Process Automation (RPA) proving its transformational value at Deutsche Bank

RPA has improved productivity and quality at Deutsche Bank. "Robots" are intended to augment the bank’s workforce, making employees happier and more efficient. At least when they are employed and working ... but what then?

What is the difference, if any, between automation of a repetitive task, process and control system and augmentation of intelligence? Why is the difference important and what is the long term impact on society of autonomous knowledge acquisition? How do behavior modification and learning automation, cognitive robotics change the paradigm and why should it matter?
Automation[1] or automatic control, is the use of various control systems for operating equipment such as machinery, processes in factories, boilers and heat treating ovens, switching on telephone networks, steering and stabilization of ships, aircraft and other applications and vehicles with minimal or reduced human intervention. Some processes have been completely automated. 
Automation has been achieved by various means including mechanical, hydraulic, pneumatic, electrical, electronic devices and computers, usually in combination. Complicated systems, such as modern factories, airplanes and ships typically use all these combined techniques. The biggest benefit of automation is that it saves labor; however, it is also used to save energy and materials and to improve quality, accuracy and precision. 
The term automation, inspired by the earlier word automatic (coming from automaton), was not widely used before 1947, when Ford established an automation department.[1] It was during this time that industry was rapidly adopting feedback controllers, which were introduced in the 1930s.[2]
Autopilot was first offered on October 9, 2014, for Tesla Model S, followed by the Model X upon its release.[5][6] Autopilot was part of a US$2,500 "Tech Package" option. At that time Autopilot features included semi-autonomous drive and parking capabilities.[7][8][9] Initial versions of Autopilot were developed in partnership with the Israeli company Mobileye.[10] Tesla and Mobileye ended their partnership in July 2016.[11][12]
Mobileye's technology is based on the use of optical vision systems with motion detection algorithms, unlike many other systems which use a combination of visual detection, radar, and laser scanning. The firm's vehicle detection algorithms recognize motorised vehicles such as cars, motorcycles and trucks, in day and night time conditions. The firm's version performs its vehicle detection based functions using a single camera mounted in the rear view mirror, unlike the usual approach of using radars, laser scanners or in some cases stereo-cameras. 
In October 2015, Tesla released software package version 7.0 with Autopilot to its customers.[13] In December 2015, Tesla announced that it will remove some self-driving features to discourage customers from engaging in risky behavior. Autopilot Firmware 7.1 made those changes and includes remote parking technology known as Summon that can park and "unpark" without the driver in the car.[14][15][16] 
On August 31, 2016, Elon Musk announced Autopilot Firmware 8.0, that processes radar signals to create a coarse point cloud similar to Lidar to help navigate in low visibility conditions, and even to 'see' in front of the car ahead.[17][18] Autopilot, as of version 8, uses radar as the primary sensor instead of the camera.[19] In November 2016, Autopilot 8.0 was updated to have a more noticeable signal that it is engaged and it requires drivers to touch the steering wheel more frequently, otherwise Autopilot will turn off.[20] By November 2016, Autopilot had operated actively on hardware version 1 vehicles for 300 million miles (500 million km) and 1.3 billion miles (2 billion km) in shadow mode.[21]As of October 2016, all Tesla vehicles come with the necessary sensing and computing hardware, known has Hardware version 2 (HW2), for future fully autonomous operation (SAE Level 5), with software being made available as it matures.[22] The company offers various free/extra-cost options for enabling Autopilot-associated features/services. 
Autopilot on hardware version 1 cars is available for US$2,500 ($3,000 after delivery) and for HW2 cars, Autopilot is available as "Enhanced Autopilot" for $5,000 ($6,000 after delivery) and additionally full self-driving capability is $3,000 ($4,000 after delivery).[23] 
The first release of Autopilot for HW2 cars was in February 2017. It included adaptive cruise control, autosteer that was enabled on divided highways, autosteer on 'local roads’ up to a speed of 35 mph or a specified number of mph over the local speed limit.[24] Firmware version 8.1 for HW2 began in June 2017 that has many new features including a new Autopilot driving-assist algorithm, full-speed braking and handling parallel and perpendicular parking.[25] 
On April 28, 2017, Elon Musk predicted that in around two years drivers would be able to sleep in their Tesla until it finishes the trip.[26] 

So far ... nothing outstanding ... autopilot systems using sensors have been with humanity for decades. 


Dean Mazboudi has been leading Deutsche Bank’s Innovation Lab in New York for two-and-a-half years, spending his days assessing and experimenting with a host of emerging technologies that could meet the company’s evolving demands. Robotic process automation (RPA) is one technology, he says, that’s proving beneficial to increasing efficiency and effectiveness for back- and mid-office processes.

[ Related: Building a business case for offshore robotic process automation ]

SHIFTING SAND OF HUMANITY, asymmetrical cognitive robotics ...does it end well?

Basic rules-based automation has been available for years, but advancing RPA tools —particularly when coupled with cognitive capabilities — are now able to transform work that’s still paper based or performed manually. “We have a high volume of manual transactions that are repetitive in nature,” Mazboudi says.

“That represents a significant challenge for us and for a lot of banks.” The company has been involved in ongoing efforts to modernize systems and processes—options like new software packages or workflow modernization tools. But those deliver “incremental change at a very large cost that takes a long time to deliver,” Mazboudi says, “leaving, in the short term, an abundance of activities where manual handling of core business processes still occurs.”

RPA can automate work in functions trade finance, cash operations, loan operations, and tax, to name a few. “In every area, we’ve experimented in we’ve achieved good value,” Mazboudi says, anywhere from 30 percent to 70 percent automation. That not only makes the bank more productive, it also improves quality and decreases risk as machines make fewer errors.

[ Related: Robotic process automation – Will you be a leader or a laggard? ]

RPA has also helped decrease the time it takes to train employees. In loan operations, for example, it can take six months for an employee to work at full capacity. What’s more, the turnover in those roles is extremely high because the work can be tedious.

“By teaching a machine that set of tasks—having that knowledge encoded through robotics and cognitive computing—that knowledge is available to humans to augment their skills and accelerate the onboarding process,” Mazboudi says. The automated system can guide employees through their day-to-day work. WELL, HERE'S THE QUESTION THAT SHOULD BE ASKED ... WHY DO WE NEED HUMANS IN THE ROLES?

“We really look at it as augmenting our workforce by making this encoded intelligence available to them,” says Mazboudi. “I don’t think robots will ever replace humans. But robots will make humans more efficient and smarter.” They could make employees happier as well. Automating more of the monotonous tasks can increase employee satisfaction, Mazboudi says. Yes, increased employee satisfaction is all well and good ... but what of the displacement within the work force ... a force multiplier using RPA impacts the social security fabric and metrics of the safety net for the mere mortals having developed the very systems purportedly employed to increased employee satisfaction. 

[ Related: Why automation doubles IT outsourcing cost savings ]

But RPA is not a quick fix. It often times requires rethinking existing business processes. “Very seldom can we take a process as it exists today and just automate it,” Mazboudi says. “If there is an opportunity to modify a process to move it closer to the digital end of the spectrum, we pursue that first.” If an existing process required that a fax is received in one location, be printed out and then sent to another location where it is entered into a computer, for example, that’s not ideal. “You have to ensure that processes are properly engineered for automation,” says Mazboudi.

It’s also important to consider RPA in the context of a rapidly shifting business and technology ecosystem. “System upgrades, underlying architectures, evolving business process—all have to be considered so that effort put into automating processes is done in conjunction with others things around it,” Mazboudi says. “Anytime you’re overlaying automation onto existing systems, you have to be careful of underlying changes and be aware of what changes may or may not break the automated processes. Once [automation] is in place, you have to manage it as core systems evolve.”

Another challenge in a massive organization like Deutsche Bank is aligning the objectives of various business functions. “If you centralize the automation effort, you can benefit from that scale, but it’s hard to take that approach because priorities and incentives can differ,” says Mazboudi.