Removing Rubble with Robots

What robots can be taught to do.

 Faster Path Planning for Rubble-Roving Robots

University of Michigan News, August 13, 2021

A new algorithm designed by University of Michigan (U-M) scientists could enhance the performance of robots traversing treacherous terrain. The algorithm can accelerate path planning for robots that use arm-like appendages to maintain balance on rough ground like disaster areas or construction sites. U-M's Yu-Chi Lin said, "First, we used machine learning to train the robot on the different ways it can place its hands and feet to maintain balance and make progress. Then, when placed in a new, complex environment, the robot can use what it learned to determine how traversable a path is, allowing it to find a path to the goal much faster." In a virtual experiment, a humanoid robot using the new algorithm outperformed previous methods in negotiating a corridor of rubble..... 

Flawed Algorithms

Always looking at data and context involved when algorithms fail.  Like any kind of predictive approach, its rarely perfect.   The same for classing analytics and for machine learning based methods.  And for that matter for human predictive methods as well.   And that can change as data and context changes over time.    So these kinds of test examples are useful.

Algorithm That Predicts Deadly Infections Is Often Flawed   By Wired

A study using data from nearly 30,000 patients in University of Michigan hospitals suggests Epic Systems' early warning system for sepsis infections performs poorly.

An algorithm designed by U.S. electronic health record provider Epic Systems to forecast sepsis infections is significantly lacking in accuracy, according to an analysis of data on about 30,000 patients in University of Michigan (U-M) hospitals.

U-M researchers said the program overlooked two-thirds of the approximately 2,500 sepsis cases in the data, rarely detected cases missed by medical staff, and was prone to false alarms.

The researchers said Epic tells customers its sepsis alert system can correctly differentiate two patients with and without sepsis with at least 76% accuracy, but they determined it was only 63% accurate.

U-M's Karandeep Singh said the study highlights wider shortcomings with proprietary algorithms increasingly used in healthcare, noting that the lack of published science on these models is "shocking."

From Wired

Worlds Smallest Computer at Work

An unusual example of embedding computing using sensors.

Snails Carrying World's Smallest Computer Help Solve Mass Extinction Survivor Mystery

University of Michigan News Service

By Katherine McAlpine; Catharine June  University of Michigan,  June 15, 2021

University of Michigan (U-M) biologists and engineers used the world's smallest computer to learn how the South Pacific Society Islands tree snail Partula hyalina survived a mass extinction. Former U-M researcher Inhee Lee adapted the Michigan Micro Mote (M3) sensor to test the theory that P. hyalina survived the deliberate introduction of the predatory rosy wolf snail to its environment as attributable to its light-reflecting white shell. The researchers deployed 50 M3s in Tahiti, gluing some to rosy wolf snails while others were stuck on leaves harboring the P. hyalina, which rests in daytime. Lee wirelessly downloaded data from each M3 at the end of the day. Based on that data, the researchers suspect P. hyalina avoids predation because the rosy wolf snail will not venture far into its sunlight-heavy habitat.

Shape-Shifting Processor for Security

Now here  is a kind of remarkable thing.   Would it work for post-quantum computing?   Could it be used to ensure all kinds of threats are addressed?  What re coding would be required to make sure of that?  Nice idea. Like the approach of large scale, multi agent testing. 

ACM NEWS

Shape-shifting Computer Chip Thwarts Hackers  By The Conversation   May 20, 2021

We have developed and tested a secure new computer processor that thwarts hackers by randomly changing its underlying structure, thus making it virtually impossible to hack.

Last summer, 525 security researchers spent three months trying to hack our Morpheus processor as well as others. All attempts against Morpheus failed. (this link also gives a technical outline of the approach) This study was part of a program sponsored by the U.S. Defense Advanced Research Program Agency to design a secure processor that could protect vulnerable software. DARPA released the results on the program to the public for the first time in January 2021.

A processor is the piece of computer hardware that runs software programs. Since a processor underlies all software systems, a secure processor has the potential to protect any software running on it from attack. Our team at the University of Michigan first developed Morpheus, a secure processor that thwarts attacks by turning the computer into a puzzle, in 2019. ...

Massive Cell Data sets

Analyzing organisms with minimal computing power. 

Algorithm Uses Online Learning for Massive Cell Datasets

Michigan Medicine, Kelly Malcolm, April 19, 2021

An algorithm developed by University of Michigan (U of M) researchers employs online learning to accelerate the analysis of enormous cell datasets, using the amount of memory found on a standard laptop computer. The algorithm enables new datasets to be added to existing ones without reprocessing the older datasets, and allows researchers to segment datasets into mini-batches so less memory is required for processing. U of M's Joshua Welch said, "Our technique allows anyone with a computer to perform analyses at the scale of an entire organism. That's really what the field is moving towards."