Boston Dynamics And Google Robots “Handle” Things Better Than You Do

Published in

SyncedReview

4 min readMar 31, 2019

Subscribe to Synced Global AI Weekly

Let Boston Dynamics ‘Handle’ That Package
Boston Dynamics introduced Handle two years ago, but the bot’s debut was overshadowed by viral videos of the company’s humanoid Atlas doing parkour and dancing robodogs Spot and SpotMini. Now Handle is a star in its own right.
(Synced)

Unifying Physics And Deep Learning With TossingBot
Google Researchers worked with researchers at Princeton, Columbia, and MIT to develop TossingBot: a picking robot for our real, random world that learns to grasp and throw objects into selected boxes outside its natural range. We find that by learning to throw, TossingBot is capable of achieving picking speeds that are twice as fast as previous systems, with twice the effective placing range.
(Google AI)

Google Is Getting Back into Robotics after Its Last Attempt Fizzled Out
Google is getting back into robotics after its previous effort fizzled out and lost its leader amid allegations of sexual misconduct, The New York Times reported. Google’s revamped robotics program will focus more on simple machines that can perform and learn tasks through machine learning.
(CNBC)

OpenAI Five Finals
OpenAI will be holding their final live event for OpenAI Five at 11:30a PT on April 13th. They will work with Dota 2 and showcase aspects of OpenAI Five which we think illustrate how humans and AI will interact in the future. It has additionally turned out to be a great avenue for helping people experience modern AI.
(OpenAI)

Technology

PyramidBox: A Context-Assisted Single Shot Face Detector
This paper proposes a novel contextassisted single shot face detector, named PyramidBox to handle the hard face detection problem. Observing the importance of the context, we improve the utilization of contextual information in the following three aspects.
(Baidu)

Imperceptible, Robust, And Targeted Adversarial Examples for Automatic Speech Recognition
Adversarial examples are inputs to machine learning models designed by an adversary to cause an incorrect output. So far, adversarial examples have been studied most extensively in the image domain. In this domain, adversarial examples can be constructed by imperceptibly modifying images to cause misclassification, and are practical in the physical world.
(UCSD & Google Brain)

CityFlow: A City-Scale Benchmark for Multi-Target Multi-Camera Vehicle Tracking And Re-Identification
Urban traffic optimization using traffic cameras as sensors is driving the need to advance state-of-the-art multitarget multi-camera (MTMC) tracking. This work introduces CityFlow, a city-scale traffic camera dataset consisting of more than 3 hours of synchronized HD videos from 40 cameras across 10 intersections, with the longest distance between two simultaneous cameras being 2.5 km.
(University of Washington & NVIDIA & San Jose State University)

You May Also Like

Hinton, LeCun & Bengio Named 2018 ACM Turing Award Laureates
The ACM (Association for Computing Machinery) this morning announced Geoffrey Hinton, Yann LeCun and Yoshua Bengio as its 2018 Turing Award winners. The trio were honoured “for conceptual and engineering breakthroughsthat have made deep neural networks a critical component of computing.”
(Synced)

Word Influence Analysis Tool ‘SyncedLeg’ Open-Sourced
SyncedLeg is a tool designed to help with that by mining influential keywords from the corpus with traffic data. A team of Synced interns developed the tool over an internal two-day Hackathon, naming it after their team “机器之腿” (“Machine’s leg” in Chinese).
(Synced)